diff --git a/crates/format/src/unpack.rs b/crates/format/src/unpack.rs index 9aa3c81b2..2b8652453 100644 --- a/crates/format/src/unpack.rs +++ b/crates/format/src/unpack.rs @@ -57,8 +57,16 @@ enum Tail<'a> { } impl<'a> Unpacked<'a> { - /// Open `bytes` (compressed or not). + /// Open `bytes` (compressed or not), with the JSON part capped at [`MAX_JSON`]. pub(crate) fn new(bytes: &'a [u8]) -> Result { + Self::within(bytes, MAX_JSON) + } + + /// Open `bytes` with the JSON part capped at `max_json` bytes instead of [`MAX_JSON`], so + /// the cap is exercised wherever the test runs: a compressed file's JSON is deserialized into + /// a document several times its size, and the cap is what keeps that off the address space + /// (#3587). + fn within(bytes: &'a [u8], max_json: u64) -> Result { if !is_compressed(bytes) { let (json, tail) = split_tail(bytes); return Ok(Self { json: Json::Borrowed(json), tail: Tail::Slice(tail) }); @@ -75,13 +83,13 @@ impl<'a> Unpacked<'a> { if n == 0 || nul.is_some_and(|z| buf.len() >= z + BLOB_MAGIC.len()) { break; } - if buf.len() as u64 > MAX_JSON + CHUNK { + if buf.len() as u64 > max_json + CHUNK { return Err(FormatError::TooBig); } } let Some(z) = nul else { // No NUL, so no marker: all of it is JSON. - if buf.len() as u64 > MAX_JSON { + if buf.len() as u64 > max_json { return Err(FormatError::TooBig); } return Ok(Self { json: Json::Owned(buf), tail: Tail::None }); @@ -92,7 +100,7 @@ impl<'a> Unpacked<'a> { && buf.get(end..).is_some_and(|r| r.starts_with(BLOB_MAGIC)) && end.checked_sub(1).and_then(|k| buf.get(k)).is_some_and(|b| matches!(b, b'}' | b']' | b'"')) { - if end as u64 > MAX_JSON { + if end as u64 > max_json { return Err(FormatError::TooBig); } let read = buf.split_off(end + BLOB_MAGIC.len()); @@ -323,6 +331,24 @@ mod tests { assert_eq!(Unpacked::new(&file).unwrap().entries(&[[0, 4096]]).unwrap(), vec![Some(Arc::new(tail.clone()))]); } + /// #3587: a compressed file's JSON is capped wherever the cap runs, so a small gzip that + /// unpacks past it is refused instead of being deserialized into a document several times + /// its size. + #[test] + fn a_documents_json_is_capped() { + // The cap itself: a quarter of the address space on wasm32, 1 GiB elsewhere. + assert_eq!(MAX_JSON, if cfg!(target_arch = "wasm32") { 256 << 20 } else { 1 << 30 }); + let mut json = br#"{"format":"vectorcraft","pad":""#.to_vec(); + json.extend(std::iter::repeat(b'x').take(4096)); + json.extend(br#""}"#); + let gz = gzip(&json); + // Within the target's cap the file opens, and the JSON reads back whole. + assert!(Unpacked::new(&gz).is_ok()); + assert_eq!(Unpacked::within(&gz, json.len() as u64).unwrap().json(), json.as_slice()); + // One byte past the cap, the whole document is refused with TooBig. + assert!(matches!(Unpacked::within(&gz, json.len() as u64 - 1), Err(FormatError::TooBig))); + } + #[test] fn entries_sharing_bytes_obey_the_limits_in_memory_too() { // Each entry within MAX_BLOB, together past MAX_BLOBS: many references to the same bytes.