Bound the BD-J label parsers, and stop a crafted disc hanging the scan

Ten defects in code no previous round had ever scoped. `src/labels/`
identifies a disc's studio by parsing jar archives and JVM class files off
untrusted media, so every byte here is attacker-controllable — and 813 of
its lines were executed by no test at all.

The worst is a non-terminating loop. A fallback stream-number scan
advanced with `saturating_add`, and the comment says why: a crafted XML
"must not overflow (panic in debug, wrap-to-0 in release)". Once the
counter pins at u16::MAX and that number is taken, the loop cannot exit.
So a fix for an overflow panic produced an unbounded hang, which is
strictly worse — a panic is observable and catchable, and catch_unwind
cannot interrupt a live loop. Reachable from about 8 MB of XML.

Where the same overflow appears in the deluxe decoder the fix is
checked_add and stop, NOT saturation — twice wrong there, because
saturating would peg every stream past the ceiling at one number and
apply_labels binds on (type, number), silently mislabelling tracks. A
correctness bug wearing the costume of success.

Round 7 capped the ldc-string retention per class; nothing capped the
aggregate, so a 64 MiB jar held that budget for every class at once. Same
defect one level up, which is the shape that keeps recurring in this
directory. Four other amplifications are bounded the same way, each with a
stated headroom and a paired test proving real media passes untouched —
the tightest is 5x on a label length, the loosest 2000x on the stream
numbering space, against BD's 32-per-type STN_table limit.

Two are not caps at all: a quadratic membership scan became a set, and an
attacker-derived length added to a cursor without saturation now cannot
wrap. Nothing is excluded by either.

A `#[cfg(test)]` hand-copy of a shipping parser was the ninth bad test
this audit has found, and the first proven by mutation rather than
inspection: deleting the guard from the REAL function left all 26 tests
green, including the one named for that guard. Pointed at the real
function, the same mutation fails.

Separately, all three failure arms of the bounded fsync returned Ok(()) on
both macOS and Linux, so sync_all reported success for a durability
barrier that never ran. Only macOS was in scope; the Linux twin is fixed
here too, because a platform disagreeing with its sibling about whether a
failed sync is an error is the class that already produced an over-length
SCSI CDB macOS rejected and the other two truncated. Note the behaviour
change: a mux whose final sync times out on a wedged mount now fails
rather than exiting 0.

Three of the caps are proven by wall-clock deadline rather than an
operation count, with 18-80x margin on the passing side. On a heavily
oversubscribed machine those could flake.

src/labels/mod.rs

@@ -977,18 +977,29 @@ pub(crate) fn jar_inventory(udf: &UdfFs) -> Vec<String> {
     let Some(jar_dir) = udf.find_dir("/BDMV/JAR") else {
         return Vec::new();
     };
-    let mut out: Vec<String> = Vec::new();
-    for entry in &jar_dir.entries {
+    jar_inventory_from(&jar_dir.entries)
+}
+
+/// The body of [`jar_inventory`], over the `/BDMV/JAR` children directly, so
+/// it is unit-testable without a `UdfFs`.
+///
+/// A `BTreeSet`, not `Vec::contains`: the entry names come from the disc's own
+/// UDF directory records, so both the file count and the name lengths are
+/// attacker-controlled, and a linear `contains` doing a full `String` compare
+/// per candidate is quadratic in the number of files. The set also subsumes
+/// the trailing sort — it yields sorted, deduplicated output directly.
+fn jar_inventory_from(entries: &[crate::udf::DirEntry]) -> Vec<String> {
+    let mut out: std::collections::BTreeSet<&str> = std::collections::BTreeSet::new();
+    for entry in entries {
         if entry.is_dir {
             for child in &entry.entries {
-                if !child.is_dir && !out.contains(&child.name) {
-                    out.push(child.name.clone());
+                if !child.is_dir {
+                    out.insert(child.name.as_str());
                 }
             }
         }
     }
-    out.sort();
-    out
+    out.into_iter().map(str::to_string).collect()
 }
 
 // ── Shared helpers ─────────────────────────────────────────────────────────
@@ -1031,6 +1042,94 @@ pub(crate) fn read_jar_file(
 mod registry_tests {
     use super::*;
 
+    fn dir_entry(
+        name: &str,
+        is_dir: bool,
+        entries: Vec<crate::udf::DirEntry>,
+    ) -> crate::udf::DirEntry {
+        crate::udf::DirEntry {
+            name: name.to_string(),
+            is_dir,
+            meta_lba: 0,
+            size: 0,
+            entries,
+        }
+    }
+
+    /// `jar_inventory` deduplicated with a linear `Vec::contains`, doing a full
+    /// `String` comparison per candidate — quadratic in a file count taken
+    /// straight from the disc's UDF directory records, with attacker-chosen
+    /// name lengths to inflate each comparison.
+    ///
+    /// Proof is by deadline. With the linear scan this fixture measures well
+    /// past the deadline; with a set it is milliseconds. Bounded so a
+    /// regression fails fast instead of hanging CI.
+    #[test]
+    fn jar_inventory_dedup_is_not_quadratic() {
+        const FILES: usize = 120_000;
+        let (tx, rx) = std::sync::mpsc::channel();
+        let worker = std::thread::spawn(move || {
+            // Long shared prefix so every comparison runs to the tail.
+            let prefix = "a".repeat(180);
+            let children: Vec<crate::udf::DirEntry> = (0..FILES)
+                .map(|i| dir_entry(&format!("{prefix}{i:08}.png"), false, Vec::new()))
+                .collect();
+            let entries = vec![dir_entry("00000", true, children)];
+            let _ = tx.send(jar_inventory_from(&entries));
+        });
+        match rx.recv_timeout(std::time::Duration::from_secs(10)) {
+            Ok(names) => {
+                worker.join().expect("worker panicked");
+                assert_eq!(names.len(), FILES);
+            }
+            Err(_) => panic!(
+                "jar_inventory_from did not finish {FILES} entries within 10s \
+                 — the dedup is still a linear scan"
+            ),
+        }
+    }
+
+    /// Behaviour contract: output is deduplicated across subdirectories,
+    /// sorted, and excludes directories and files sitting directly under
+    /// `/BDMV/JAR` (only one level down counts).
+    #[test]
+    fn jar_inventory_dedups_sorts_and_skips_dirs() {
+        let entries = vec![
+            dir_entry(
+                "00000",
+                true,
+                vec![
+                    dir_entry("streamproperties.xml", false, Vec::new()),
+                    dir_entry("zeta.png", false, Vec::new()),
+                    dir_entry(
+                        "nested",
+                        true,
+                        vec![dir_entry("hidden.txt", false, Vec::new())],
+                    ),
+                ],
+            ),
+            dir_entry(
+                "00001",
+                true,
+                vec![
+                    dir_entry("alpha.png", false, Vec::new()),
+                    // Duplicate of the entry in 00000 — must appear once.
+                    dir_entry("streamproperties.xml", false, Vec::new()),
+                ],
+            ),
+            // A jar sitting directly under /BDMV/JAR is not inventoried.
+            dir_entry("top.jar", false, Vec::new()),
+        ];
+        assert_eq!(
+            jar_inventory_from(&entries),
+            vec![
+                "alpha.png".to_string(),
+                "streamproperties.xml".to_string(),
+                "zeta.png".to_string(),
+            ]
+        );
+    }
+
     /// Lock the parser roster + order. If someone reorders the array
     /// or adds/removes a parser, this test forces them to update the
     /// expectation explicitly. The order is load-bearing: first