diagnose native Recovery disk IPC with bounded observation

tools/ci/macos-native-readiness.sh

@@ -14,6 +14,9 @@ TIMER_FIFO="/tmp/native-diagnostic-$PROOF_TOKEN-$$.fifo"
 PENDING_OUTPUTS=()
 ACTIVE_COMMAND=""
 ACTIVE_TIMER=""
+# BEGIN disk IPC diagnostic
+ACTIVE_OBSERVER=""
+# END disk IPC diagnostic
 os_version=""
 architecture=""
 uid=-1
@@ -96,8 +99,75 @@ read_scalar() {
   SCALAR="$value"
 }
 
+# BEGIN disk IPC diagnostic
+# Optional observations have their own child/timer ownership. No service is
+# loaded, restarted or changed, and samples target only this probe's diskutil.
+observe_disk_query() {
+  local disk_process="$1" output="$2" sample_output="$3" observation_child="" observation_timer=""
+  cancel_observation() {
+    trap '' TERM INT
+    if [ -n "$observation_child" ]; then
+      kill -TERM "$observation_child" 2>/dev/null || :
+      IFS= read -r -t 2 -u 9 unused || :
+      kill -KILL "$observation_child" 2>/dev/null || :
+      wait "$observation_child" 2>/dev/null || :
+    fi
+    [ -z "$observation_timer" ] || { kill -TERM "$observation_timer" 2>/dev/null || :; wait "$observation_timer" 2>/dev/null || :; }
+    printf '[disk-observation] stopped after the owned disk query\n' >> "$output"
+    exit 143
+  }
+  observe_command() {
+    local name="$1" status started=$SECONDS
+    shift
+    printf '\n[disk-observation-command] %s:' "$name" >> "$output"
+    printf ' %s' "$@" >> "$output"
+    printf '\n' >> "$output"
+    "$@" >> "$output" 2>&1 &
+    observation_child=$!
+    (
+      trap 'exit 0' TERM INT
+      IFS= read -r -t 8 -u 9 unused || :
+      printf '[disk-observation-timeout] %s child=%s limit=8s\n' "$name" "$observation_child" >> "$output"
+      kill -TERM "$observation_child" 2>/dev/null || :
+      IFS= read -r -t 2 -u 9 unused || :
+      kill -KILL "$observation_child" 2>/dev/null || :
+    ) &
+    observation_timer=$!
+    wait "$observation_child"; status=$?
+    kill -TERM "$observation_timer" 2>/dev/null || :
+    wait "$observation_timer" 2>/dev/null || :
+    printf '[disk-observation-exit] %s status=%s elapsed=%ss\n' "$name" "$status" "$((SECONDS - started))" >> "$output"
+    observation_child=""; observation_timer=""
+  }
+  trap cancel_observation TERM INT
+  printf '[disk-observation] owned-diskutil-child=%s parent-shell=%s\n' "$disk_process" "$$" >> "$output"
+  observe_command processes /bin/ps -axo pid,ppid,state,comm
+  if [ -x /usr/bin/sample ]; then
+    if kill -0 "$disk_process" 2>/dev/null; then
+      observe_command diskutil-sample /usr/bin/sample "$disk_process" 3 10 -file "$sample_output"
+    else
+      printf '[disk-observation-unavailable] diskutil already exited before sample\n' >> "$output"
+    fi
+  else
+    printf '[disk-observation-unavailable] /usr/bin/sample is unavailable\n' >> "$output"
+  fi
+  observe_command arbitration /bin/launchctl print system/com.apple.diskarbitrationd
+  observe_command management /bin/launchctl print system/com.apple.diskmanagementd
+}
+
+stop_disk_observation() {
+  [ -n "$ACTIVE_OBSERVER" ] || return 0
+  kill -TERM "$ACTIVE_OBSERVER" 2>/dev/null || :
+  wait "$ACTIVE_OBSERVER" 2>/dev/null || :
+  ACTIVE_OBSERVER=""
+}
+
+# END disk IPC diagnostic
 cancel_probe() {
   trap '' TERM INT
+# BEGIN disk IPC diagnostic
+  stop_disk_observation
+# END disk IPC diagnostic
   if [ -n "$ACTIVE_COMMAND" ]; then
     kill -TERM "$ACTIVE_COMMAND" 2>/dev/null || :
     IFS= read -r -t 2 -u 9 unused || :
@@ -136,6 +206,18 @@ run_command() {
   ) &
   timer=$!
   ACTIVE_TIMER="$timer"
+# BEGIN disk IPC diagnostic
+  if [[ "$name" == disks && "$attempt" == 1 ]]; then
+    local observation_output="/tmp/native-diagnostic-disk-observation.out"
+    local sample_output="/tmp/native-diagnostic-disk-sample.out"
+    : > "$observation_output"
+    : > "$sample_output"
+    observe_disk_query "$process" "$observation_output" "$sample_output" &
+    ACTIVE_OBSERVER=$!
+    printf '[disk-observation-start] observer=%s owned-diskutil-child=%s\n' "$ACTIVE_OBSERVER" "$process" >&3
+    PENDING_OUTPUTS+=("$observation_output" "$sample_output")
+  fi
+# END disk IPC diagnostic
   wait "$process"
   exit_code=$?
   waited=$SECONDS
@@ -143,6 +225,9 @@ run_command() {
   printf '[proof-native-wait] %s child=%s elapsed=%ss exit=%s\n' "$name" "$process" "$((waited - started))" "$exit_code" >&3
   kill -TERM "$timer" 2>/dev/null || :
   wait "$timer" 2>/dev/null || :
+# BEGIN disk IPC diagnostic
+  stop_disk_observation
+# END disk IPC diagnostic
   ACTIVE_COMMAND=""; ACTIVE_TIMER=""
   printf '[proof-cleanup] %s child=%s elapsed=%ss total=%ss\n' "$name" "$process" "$((SECONDS - waited))" "$((SECONDS - started))" >&3
   printf '[proof-exit] %s\n' "$exit_code" >&3
@@ -203,10 +288,17 @@ os_version="$SCALAR"
 (( ${os_version%%.*} >= 13 )) || fail_probe unsupported_macos_version
 flush_outputs || finish false diagnostic_log_budget_exceeded
 
+# BEGIN disk IPC diagnostic
+run_command arbitration_before /bin/launchctl print system/com.apple.diskarbitrationd
+run_command management_before /bin/launchctl print system/com.apple.diskmanagementd
+run_command media_before /usr/sbin/ioreg -r -c IOMedia -l -w 0
+flush_outputs || finish false diagnostic_log_budget_exceeded
+
+# END disk IPC diagnostic
 # Bound readiness independently of the host's 40-minute overall deadline.
 readiness_start=$SECONDS
 attempt=0
-while (( SECONDS - readiness_start < 600 )); do
+while (( attempt < 2 && SECONDS - readiness_start < 600 )); do
   attempt=$((attempt + 1))
   printf '\n[readiness-attempt] %s\n' "$attempt" >&3
   run_command disks /usr/sbin/diskutil list physical