Skip to content

perf(ffe)!: cache the Python object for each JSON variant - #2889

Draft
leoromanovsky wants to merge 1 commit into
leo.romanovsky/ffe-json-scale-benchfrom
leo.romanovsky/ffe-json-pyo3-cache
Draft

leoromanovsky wants to merge 1 commit into
leo.romanovsky/ffe-json-scale-benchfrom
leo.romanovsky/ffe-json-pyo3-cache

Conversation

@leoromanovsky

@leoromanovsky leoromanovsky commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

This PR is stacked on #2888 (scale-suite benchmark and conformance test). Review only the commits after it: 478310f.

What does this PR do?

Caches the Python object that the pyo3 binding builds for each JSON variant, so that repeat evaluations of a JSON flag return the same object instead of rebuilding the tree.

  • AssignmentValue::Json gains a cache: JsonValueCache field. With the pyo3 feature, the cache holds an Arc<PyOnceLock<Py<PyAny>>>. Without it, JsonValueCache is a zero-size type, so the Rust core and the C FFI are unchanged.
  • Each split clones its variation's AssignmentValue, so all assignments of one variant share one cell. The cache lives exactly as long as the Configuration. There is no global cache, no expiry, and no key by flag name. A new configuration builds new cells, so the first evaluation after a configuration update converts again.
  • IntoPyObject for &AssignmentValue fills the cell on first use with the existing json_to_pyobject conversion and returns the cached object on later calls. The number conversion (u128/i128/f64) is unchanged.

Motivation

A performance-sensitive customer serves large JSON flag values from Python. The core already shares each parsed JSON value (#2888 measures 120–145 ns with no allocation for 719 nodes up to 4 MB), but every Python read rebuilt the whole object tree: about 106 µs for 719 nodes and 63 ms for 4 MB, on every evaluation.

flowchart LR
  subgraph Before
    C1[Configuration<br/>Arc&lt;serde_json::Value&gt;] -->|every read| J1[json_to_pyobject] --> P1[new dict tree]
  end
  subgraph After
    C2[Configuration<br/>Arc&lt;serde_json::Value&gt;] -->|first read| J2[json_to_pyobject] --> K[PyOnceLock per variant]
    K -->|every read| P2[same dict]
  end
Loading

Concurrency. PyOnceLock detaches from the interpreter while it waits for another thread to finish initializing, so a thread that holds the GIL cannot deadlock against the initializer. On free-threaded builds it is a real once-lock, so concurrent first reads produce exactly one object. If two threads race, one waits; neither sees a partly built object.

Behavior changes

  • Repeat evaluations of a JSON flag return the same Python object until the next configuration update. Callers must treat it as read-only: a caller that mutates it changes what later evaluations return. dd-trace-py documents this in its stacked PR.
  • Breaking: AssignmentValue::Json has a new public field, so code that constructs or exhaustively destructures it must add cache (or ..). In this repository only libdd-ffe-ffi needed a change.
  • The cached object is kept alive by the configuration, so a configuration's memory now includes one Python tree per JSON variant that has been read.

Results

Measured through dd-trace-py's provider (DataDogProvider.resolve_object_details) with the openfeature_scale_json benchmark from DataDog/dd-trace-py#21875. The before and after builds come from the same cargo command and differ only in the libdd-ffe revision (v49.0.0 vs this PR). Each pair ran back to back (after, before, after, before) on one 16 vCPU workspace, Python 3.12, pyperf --fast median. Load averages at the start of each run: 3.06, 3.25, 2.51, 2.88. The load rose to 16 near the end of the last run, which explains the wide spread on its last 4 MB case.

Repeat evaluation, same configuration:

Case Before (run 1 / run 2) After (run 1 / run 2) Peak memory per eval, before → after
719 nodes 105.9 / 107.3 µs 4.96 / 4.94 µs 62 KB → 0.8 KB
100 KB 1042 / 1058 µs 5.43 / 5.50 µs 686 KB → 0.8 KB
1 MB 10.89 / 11.19 ms 5.39 / 5.37 µs 7.06 MB → 0.8 KB
4 MB 62.75 / 63.73 ms 5.41 / 5.38 µs 28.2 MB → 0.8 KB

First evaluation after a configuration load (the cache is empty, so it still converts):

Case Before (run 1 / run 2) After (run 1 / run 2)
719 nodes 141.9 / 137.5 µs 126.6 / 128.5 µs
100 KB 1224 / 1207 µs 1048 / 1054 µs
1 MB 12.30 / 12.49 ms 10.47 / 10.31 ms
4 MB 63.35 / 67.29 ms 52.64 / 52.44 ms

The first evaluation is 8–17% faster even though it does the same conversion. I have not isolated why. A likely cause is that the tree is now freed with the configuration, outside the timed call, rather than when the previous result is dropped. Peak memory for the first evaluation is unchanged.

The Rust core numbers in #2888 do not change: without pyo3, the cache is a zero-size field.

No number-precision bugs were found in these suites.

Additional Notes

How to test the change?

cargo test -p libdd-ffe --features pyo3 --lib pyo3_tests

repeat_evaluations_share_one_python_object checks that two evaluations of the same variant return the same object and that a new configuration returns a different one. cargo nextest run --workspace passes (178 tests), and clippy is clean with and without --features pyo3.

🤖 Generated with Claude Code

The pyo3 conversion of `AssignmentValue::Json` built a new `dict` and
`list` tree on every evaluation, and on every read of the value. That
costs milliseconds per evaluation for megabyte values.

`AssignmentValue::Json` now carries a `JsonValueCache` that every clone
shares with the variation in the configuration. The pyo3 binding fills
it on first use through `PyOnceLock`, which detaches from the
interpreter while it waits and so cannot deadlock with the GIL, and
which is thread-safe on free-threaded builds. Each repeat evaluation of
a variation returns the same object. The cache is dropped with the
configuration, and a new configuration starts empty.

Callers must treat the returned object as read-only: a change is
visible to every later evaluation of that variation.

BREAKING CHANGE: `AssignmentValue::Json` has a new `cache` field.
Patterns must use `..`, and constructors set
`cache: JsonValueCache::default()`.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

📚 Documentation Check Results

⚠️ 897 documentation warning(s) found

📦 libdd-ffe-ffi - 534 warning(s)

📦 libdd-ffe - 363 warning(s)


Updated: 2026-11-18 02:24:43 UTC | Commit: bc917b3 | missing-docs job results

@github-actions

Copy link
Copy Markdown
Contributor

🔒 Cargo Deny Results

⚠️ 15 issue(s) found, showing only errors (advisories, bans, sources)

📦 libdd-ffe-ffi - 7 error(s)

Show output
info: syncing channel updates for 1.91.1-x86_64-unknown-linux-gnu
info: latest update on 2025-11-10 for version 1.91.1 (ed61e7d7e 2025-11-07)
info: downloading 6 components
warn: the missing active toolchain `1.91.1-x86_64-unknown-linux-gnu` has been auto-installed
warn: this might cause rustup commands to take longer time to finish than expected
info: you may opt out with `RUSTUP_AUTO_INSTALL=0` or `rustup set auto-install disable`
error[unmaintained]: Bincode is unmaintained
   ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:25:1
   │
25 │ bincode 1.3.3 registry+https://github.com/rust-lang/crates.io-index
   │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unmaintained advisory detected
   │
   ├ ID: RUSTSEC-2025-0141
   ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2025-0141
   ├ Due to a doxxing and harassment incident, the bincode team has taken the decision to cease development permanently.
     
     The team considers version 1.3.3 a complete version of bincode that is not in need of any updates.
     
     ## Alternatives to consider
     
     * [wincode](https://crates.io/crates/wincode)
     * [postcard](https://crates.io/crates/postcard)
     * [bitcode](https://crates.io/crates/bitcode)
     * [rkyv](https://crates.io/crates/rkyv)
   ├ Announcement: https://git.sr.ht/~stygianentity/bincode/tree/v3.0/item/README.md
   ├ Solution: No safe upgrade is available!
   ├ bincode v1.3.3
     └── (dev) libdd-ffe v2.0.0
         └── libdd-ffe-ffi v1.0.2

error[vulnerability]: TLS 1.3 handshake messages incorrectly accepted across encryption level boundaries
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:201:1
    │
201 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0285
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0285
    ├ Rustls accepted TLS 1.3 handshake messages sent at the wrong encryption level
      when they followed a key-changing message in the same record. For example,
      a plaintext `EncryptedExtensions` message packed into the same record as the
      `ServerHello` was accepted.
      
      RFC 8446 section 5.1 requires that handshake messages do not span key changes,
      and that implementations terminate the connection with an "unexpected_message"
      alert if they do.
      
      The handshake transcript is still authenticated, so a network-position attacker
      cannot use this to alter or complete a handshake; the practical effect is that
      a peer could send handshake messages that should be encrypted in plaintext
      without rustls rejecting the connection.
      
      This is functionally the same bug as Go's
      [GO-2026-4340](https://pkg.go.dev/vuln/GO-2026-4340) (CVE-2025-61730).
    ├ Announcement: https://github.com/rustls/rustls/pull/3265
    ├ Solution: Upgrade to >=0.23.45 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   │   └── libdd-ffe-ffi v1.0.2
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[vulnerability]: `Debug` impls for client session values expose TLS secrets
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:201:1
    │
201 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0346
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0346
    ├ The `Debug` implementations of `Tls12ClientSessionValue` and
      `Tls13ClientSessionValue` were derived, and so included the TLS 1.2 master
      secret or TLS 1.3 resumption PSK in their output.
      
      Custom `ClientSessionStore` implementations that format these values with
      `Debug` (for example, when logging) could disclose these secrets to readers of
      the logs. This conflicts with rustls's
      [key-material redaction policy](https://github.com/rustls/rustls/blob/99f2358cae2954837dbb866faf6727de75489ab9/SECURITY.md#boundary-public-api).
      
      Ordinary connection and built-in session cache `Debug` output is redacted and
      is not affected.
    ├ Announcement: https://github.com/rustls/rustls/pull/3299
    ├ Solution: Upgrade to >=0.23.46 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   │   └── libdd-ffe-ffi v1.0.2
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[vulnerability]: Brotli certificate decompression can allocate up to 1 GiB before authentication
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:201:1
    │
201 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0348
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0348
    ├ The built-in Brotli certificate decompressor (enabled by the non-default
      `brotli` crate feature) bounded the size of the decompressed output, but
      accepted a peer-selected history window. Brotli's large-window extension was
      enabled, so a five-byte header could cause an allocation of 1 GiB. A peer can
      abort a process whose allocator cannot satisfy that request.
      
      No trusted certificate is required by the attacker. Affected configurations are:
      
      - clients that enable the `brotli` feature and connect to a malicious server, and
      - servers that enable the `brotli` feature and request client certificates.
      
      Applications that do not enable the `brotli` feature are not affected.
    ├ Announcement: https://github.com/rustls/rustls/pull/3297
    ├ Solution: Upgrade to >=0.23.46 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   │   └── libdd-ffe-ffi v1.0.2
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[vulnerability]: Unauthenticated ServerHello can downgrade later connections to classical key exchange
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:201:1
    │
201 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0349
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0349
    ├ The client recorded the server-selected key exchange group for each server
      name when processing the ServerHello, before the server was authenticated.
      This record is used to predict which key share to send in future ClientHellos
      to that server name.
      
      A network attacker can answer one connection attempt with a ServerHello
      selecting X25519, and then close the connection. Later ClientHellos for that
      name carry only an X25519 key share, while `supported_groups` still lists
      X25519MLKEM768 first. If the genuine server accepts that share rather than
      sending a HelloRetryRequest for its preferred hybrid group, the handshake
      completes with classical-only key exchange and no error. This persists for as
      long as the entry stays in the client's session store.
      
      The impact is loss of post-quantum confidentiality for affected connections
      ("harvest now, decrypt later"). There is no key compromise and no
      authentication bypass.
      
      The following are not affected:
      
      - connections to servers which send a HelloRetryRequest for their preferred group,
      - clients with resumption disabled (as nothing is stored), and
      - other server names.
      
      This behaviour is discussed in
      [draft-ietf-tls-key-share-prediction](https://www.ietf.org/archive/id/draft-ietf-tls-key-share-prediction-01.html#name-security-considerations),
      so the practical impact depends largely on server policy.
    ├ Announcement: https://github.com/rustls/rustls/pull/3355
    ├ Solution: Upgrade to >=0.23.46 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   │   └── libdd-ffe-ffi v1.0.2
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[unsound]: `AsyncFd::new` and related constructors violate I/O safety
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:252:1
    │
252 │ tokio 1.49.0 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unsound advisory detected
    │
    ├ ID: RUSTSEC-2026-0386
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0386
    ├ `tokio::io::unix::AsyncFd` provides safe constructors (`new`, `with_interest`, `try_new`, and `try_with_interest`) that accept any `T: AsRawFd` rather than requiring proof of ownership or validity of the file descriptor. Because `RawFd` implements `AsRawFd`, safe code can construct an `AsyncFd` from an arbitrary or closed file descriptor, or close/replace the underlying file descriptor while the `AsyncFd` is still alive.
      
      This violates I/O safety in two ways:
      
      1. `AsyncFd<T>` implements `AsFd` for all `T: AsRawFd` by calling `BorrowedFd::borrow_raw(self.as_raw_fd())`, allowing safe code to obtain an arbitrary `BorrowedFd<'_>` without owning the file descriptor.
      2. Tokio's I/O driver stores a pointer to an internal allocation in the `epoll`/`kqueue` user data for registered file descriptors. If the underlying file descriptor is duplicated and then closed while an `AsyncFd` is alive, and the OS reuses that file descriptor number for a another IO resource on the same runtime, dropping the first `AsyncFd` deregisters the reused file descriptor number and frees the first while the original file description remains registered in `epoll`, leading to a use-after-free when a readiness event fires.
      
      Starting in Tokio 1.51.6, 1.53.3, and 1.54.0, `AsyncFd::new`, `AsyncFd::with_interest`, `AsyncFd::try_new`, and `AsyncFd::try_with_interest` are deprecated in favor of the `unsafe` constructors `AsyncFd::register` and `AsyncFd::register_with_interest`. Callers must ensure that the file descriptor remains valid, open, and unchanged for the lifetime of the `AsyncFd`.
    ├ Announcement: https://github.com/tokio-rs/tokio/pull/8587
    ├ Solution: Upgrade to >=1.51.6, <1.52.0 OR >=1.53.3 (try `cargo update -p tokio`)
    ├ tokio v1.49.0
      ├── h2 v0.4.19
      │   └── hyper v1.11.1
      │       ├── httpmock v0.8.0-alpha.1
      │       │   └── (dev) libdd-ffe v2.0.0
      │       │       └── libdd-ffe-ffi v1.0.2
      │       ├── hyper-rustls v0.27.7
      │       │   └── libdd-networking v1.0.0
      │       │       ├── libdd-capabilities-impl v8.0.0
      │       │       │   ├── (dev) libdd-ffe v2.0.0 (*)
      │       │       │   └── libdd-remote-config v8.0.0
      │       │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── libdd-remote-config v8.0.0 (*)
      │       ├── hyper-util v0.1.17
      │       │   ├── httpmock v0.8.0-alpha.1 (*)
      │       │   ├── hyper-rustls v0.27.7 (*)
      │       │   ├── libdd-networking v1.0.0 (*)
      │       │   └── libdd-remote-config v8.0.0 (*)
      │       ├── libdd-common-ffi v49.0.0
      │       │   └── libdd-ffe-ffi v1.0.2 (*)
      │       ├── libdd-networking v1.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── httpmock v0.8.0-alpha.1 (*)
      ├── hyper v1.11.1 (*)
      ├── hyper-rustls v0.27.7 (*)
      ├── hyper-util v0.1.17 (*)
      ├── (dev) libdd-capabilities-impl v8.0.0 (*)
      ├── (dev) libdd-ffe v2.0.0 (*)
      ├── (dev) libdd-networking v1.0.0 (*)
      ├── libdd-remote-config v8.0.0 (*)
      ├── (dev) libdd-trace-protobuf v5.1.1
      │   └── libdd-remote-config v8.0.0 (*)
      ├── tokio-rustls v0.26.0
      │   ├── hyper-rustls v0.27.7 (*)
      │   └── (dev) libdd-networking v1.0.0 (*)
      ├── tokio-util v0.7.12
      │   ├── h2 v0.4.19 (*)
      │   └── libdd-remote-config v8.0.0 (*)
      └── (dev) tools v49.0.0
          └── (dev) libdd-networking v1.0.0 (*)

error[unmaintained]: chrono is unmaintained
   ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:48:1
   │
48 │ chrono 0.4.38 registry+https://github.com/rust-lang/crates.io-index
   │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unmaintained advisory detected
   │
   ├ ID: RUSTSEC-2026-0543
   ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0543
   ├ The chrono crate is unmaintained; all versions are affected.
     
     Recommended alternatives:
     
     - [jiff](https://crates.io/crates/jiff)
     - [time](https://crates.io/crates/time)
   ├ Announcement: https://github.com/chronotope/chrono/issues/1768
   ├ Solution: No safe upgrade is available!
   ├ chrono v0.4.38
     ├── libdd-common-ffi v49.0.0
     │   └── libdd-ffe-ffi v1.0.2
     └── libdd-ffe v2.0.0
         └── libdd-ffe-ffi v1.0.2 (*)

advisories FAILED, bans ok, sources ok

📦 libdd-ffe - 8 error(s)

Show output
error[unmaintained]: Bincode is unmaintained
   ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:23:1
   │
23 │ bincode 1.3.3 registry+https://github.com/rust-lang/crates.io-index
   │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unmaintained advisory detected
   │
   ├ ID: RUSTSEC-2025-0141
   ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2025-0141
   ├ Due to a doxxing and harassment incident, the bincode team has taken the decision to cease development permanently.
     
     The team considers version 1.3.3 a complete version of bincode that is not in need of any updates.
     
     ## Alternatives to consider
     
     * [wincode](https://crates.io/crates/wincode)
     * [postcard](https://crates.io/crates/postcard)
     * [bitcode](https://crates.io/crates/bitcode)
     * [rkyv](https://crates.io/crates/rkyv)
   ├ Announcement: https://git.sr.ht/~stygianentity/bincode/tree/v3.0/item/README.md
   ├ Solution: No safe upgrade is available!
   ├ bincode v1.3.3
     └── (dev) libdd-ffe v2.0.0

error[unsound]: Potential use-after-free due to lack of panic safety in `LruCache::pop()`
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:137:1
    │
137 │ lru 0.16.4 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unsound advisory detected
    │
    ├ ID: RUSTSEC-2026-0253
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0253
    ├ `LruCache::pop()` in `lru` was not panic-safe. If the `Drop` implementation of a stored key panics during `pop()`, `self.detach()` is never called, leaving dangling pointers in the internal doubly-linked list.
      
      A subsequent cache operation that triggers eviction can then dereference these dangling pointers:
      - The node is freed from the map, but remains linked in the LRU list due to the skipped `detach()` call
      - When a new insertion causes eviction, the LRU traversal encounters the dangling pointer
      - This results in a write to already-freed memory during the eviction process
      
      ## Impact
      
      - **CWE-416 (Use-After-Free):** memory corruption when subsequent cache operations access freed node pointers in the linked list
      - **CWE-415 (Double Free):** potential heap corruption when the same memory is freed multiple times
      
      Both types of undefined behavior can be invoked in safe Rust, but only if unwinding panics are enabled and `std::panic::catch_unwind` is used with key types that have potentially-panicking `Drop` implementations.
      
      ## Fix
      
      Fixed in `lru` 0.18.2 by detaching the node from the linked list before freeing it and dropping the key ([lru-rs#238](https://github.com/jeromefroe/lru-rs/pull/238)).
    ├ Announcement: https://github.com/jeromefroe/lru-rs/pull/238
    ├ 1 dependent has version requirements that preclude updating:
    ├ libdd-ffe = 2.0.0
    ├   - lru = '^0.16.3'
    ├ lru v0.16.4
      └── libdd-ffe v2.0.0

error[vulnerability]: TLS 1.3 handshake messages incorrectly accepted across encryption level boundaries
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:182:1
    │
182 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0285
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0285
    ├ Rustls accepted TLS 1.3 handshake messages sent at the wrong encryption level
      when they followed a key-changing message in the same record. For example,
      a plaintext `EncryptedExtensions` message packed into the same record as the
      `ServerHello` was accepted.
      
      RFC 8446 section 5.1 requires that handshake messages do not span key changes,
      and that implementations terminate the connection with an "unexpected_message"
      alert if they do.
      
      The handshake transcript is still authenticated, so a network-position attacker
      cannot use this to alter or complete a handshake; the practical effect is that
      a peer could send handshake messages that should be encrypted in plaintext
      without rustls rejecting the connection.
      
      This is functionally the same bug as Go's
      [GO-2026-4340](https://pkg.go.dev/vuln/GO-2026-4340) (CVE-2025-61730).
    ├ Announcement: https://github.com/rustls/rustls/pull/3265
    ├ Solution: Upgrade to >=0.23.45 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[vulnerability]: `Debug` impls for client session values expose TLS secrets
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:182:1
    │
182 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0346
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0346
    ├ The `Debug` implementations of `Tls12ClientSessionValue` and
      `Tls13ClientSessionValue` were derived, and so included the TLS 1.2 master
      secret or TLS 1.3 resumption PSK in their output.
      
      Custom `ClientSessionStore` implementations that format these values with
      `Debug` (for example, when logging) could disclose these secrets to readers of
      the logs. This conflicts with rustls's
      [key-material redaction policy](https://github.com/rustls/rustls/blob/99f2358cae2954837dbb866faf6727de75489ab9/SECURITY.md#boundary-public-api).
      
      Ordinary connection and built-in session cache `Debug` output is redacted and
      is not affected.
    ├ Announcement: https://github.com/rustls/rustls/pull/3299
    ├ Solution: Upgrade to >=0.23.46 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[vulnerability]: Brotli certificate decompression can allocate up to 1 GiB before authentication
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:182:1
    │
182 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0348
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0348
    ├ The built-in Brotli certificate decompressor (enabled by the non-default
      `brotli` crate feature) bounded the size of the decompressed output, but
      accepted a peer-selected history window. Brotli's large-window extension was
      enabled, so a five-byte header could cause an allocation of 1 GiB. A peer can
      abort a process whose allocator cannot satisfy that request.
      
      No trusted certificate is required by the attacker. Affected configurations are:
      
      - clients that enable the `brotli` feature and connect to a malicious server, and
      - servers that enable the `brotli` feature and request client certificates.
      
      Applications that do not enable the `brotli` feature are not affected.
    ├ Announcement: https://github.com/rustls/rustls/pull/3297
    ├ Solution: Upgrade to >=0.23.46 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[vulnerability]: Unauthenticated ServerHello can downgrade later connections to classical key exchange
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:182:1
    │
182 │ rustls 0.23.37 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ security vulnerability detected
    │
    ├ ID: RUSTSEC-2026-0349
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0349
    ├ The client recorded the server-selected key exchange group for each server
      name when processing the ServerHello, before the server was authenticated.
      This record is used to predict which key share to send in future ClientHellos
      to that server name.
      
      A network attacker can answer one connection attempt with a ServerHello
      selecting X25519, and then close the connection. Later ClientHellos for that
      name carry only an X25519 key share, while `supported_groups` still lists
      X25519MLKEM768 first. If the genuine server accepts that share rather than
      sending a HelloRetryRequest for its preferred hybrid group, the handshake
      completes with classical-only key exchange and no error. This persists for as
      long as the entry stays in the client's session store.
      
      The impact is loss of post-quantum confidentiality for affected connections
      ("harvest now, decrypt later"). There is no key compromise and no
      authentication bypass.
      
      The following are not affected:
      
      - connections to servers which send a HelloRetryRequest for their preferred group,
      - clients with resumption disabled (as nothing is stored), and
      - other server names.
      
      This behaviour is discussed in
      [draft-ietf-tls-key-share-prediction](https://www.ietf.org/archive/id/draft-ietf-tls-key-share-prediction-01.html#name-security-considerations),
      so the practical impact depends largely on server policy.
    ├ Announcement: https://github.com/rustls/rustls/pull/3355
    ├ Solution: Upgrade to >=0.23.46 (try `cargo update -p rustls`)
    ├ rustls v0.23.37
      ├── hyper-rustls v0.27.7
      │   └── libdd-networking v1.0.0
      │       ├── libdd-capabilities-impl v8.0.0
      │       │   ├── (dev) libdd-ffe v2.0.0
      │       │   └── libdd-remote-config v8.0.0
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── libdd-networking v1.0.0 (*)
      ├── rustls-platform-verifier v0.6.2
      │   └── libdd-networking v1.0.0 (*)
      └── tokio-rustls v0.26.0
          ├── hyper-rustls v0.27.7 (*)
          └── (dev) libdd-networking v1.0.0 (*)

error[unsound]: `AsyncFd::new` and related constructors violate I/O safety
    ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:232:1
    │
232 │ tokio 1.49.0 registry+https://github.com/rust-lang/crates.io-index
    │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unsound advisory detected
    │
    ├ ID: RUSTSEC-2026-0386
    ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0386
    ├ `tokio::io::unix::AsyncFd` provides safe constructors (`new`, `with_interest`, `try_new`, and `try_with_interest`) that accept any `T: AsRawFd` rather than requiring proof of ownership or validity of the file descriptor. Because `RawFd` implements `AsRawFd`, safe code can construct an `AsyncFd` from an arbitrary or closed file descriptor, or close/replace the underlying file descriptor while the `AsyncFd` is still alive.
      
      This violates I/O safety in two ways:
      
      1. `AsyncFd<T>` implements `AsFd` for all `T: AsRawFd` by calling `BorrowedFd::borrow_raw(self.as_raw_fd())`, allowing safe code to obtain an arbitrary `BorrowedFd<'_>` without owning the file descriptor.
      2. Tokio's I/O driver stores a pointer to an internal allocation in the `epoll`/`kqueue` user data for registered file descriptors. If the underlying file descriptor is duplicated and then closed while an `AsyncFd` is alive, and the OS reuses that file descriptor number for a another IO resource on the same runtime, dropping the first `AsyncFd` deregisters the reused file descriptor number and frees the first while the original file description remains registered in `epoll`, leading to a use-after-free when a readiness event fires.
      
      Starting in Tokio 1.51.6, 1.53.3, and 1.54.0, `AsyncFd::new`, `AsyncFd::with_interest`, `AsyncFd::try_new`, and `AsyncFd::try_with_interest` are deprecated in favor of the `unsafe` constructors `AsyncFd::register` and `AsyncFd::register_with_interest`. Callers must ensure that the file descriptor remains valid, open, and unchanged for the lifetime of the `AsyncFd`.
    ├ Announcement: https://github.com/tokio-rs/tokio/pull/8587
    ├ Solution: Upgrade to >=1.51.6, <1.52.0 OR >=1.53.3 (try `cargo update -p tokio`)
    ├ tokio v1.49.0
      ├── h2 v0.4.19
      │   └── hyper v1.11.1
      │       ├── httpmock v0.8.0-alpha.1
      │       │   └── (dev) libdd-ffe v2.0.0
      │       ├── hyper-rustls v0.27.7
      │       │   └── libdd-networking v1.0.0
      │       │       ├── libdd-capabilities-impl v8.0.0
      │       │       │   ├── (dev) libdd-ffe v2.0.0 (*)
      │       │       │   └── libdd-remote-config v8.0.0
      │       │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       │       └── (dev) libdd-remote-config v8.0.0 (*)
      │       │       ├── libdd-ffe v2.0.0 (*)
      │       │       └── libdd-remote-config v8.0.0 (*)
      │       ├── hyper-util v0.1.17
      │       │   ├── httpmock v0.8.0-alpha.1 (*)
      │       │   ├── hyper-rustls v0.27.7 (*)
      │       │   ├── libdd-networking v1.0.0 (*)
      │       │   └── libdd-remote-config v8.0.0 (*)
      │       ├── libdd-ffe v2.0.0 (*)
      │       ├── libdd-networking v1.0.0 (*)
      │       └── libdd-remote-config v8.0.0 (*)
      ├── httpmock v0.8.0-alpha.1 (*)
      ├── hyper v1.11.1 (*)
      ├── hyper-rustls v0.27.7 (*)
      ├── hyper-util v0.1.17 (*)
      ├── (dev) libdd-capabilities-impl v8.0.0 (*)
      ├── libdd-ffe v2.0.0 (*)
      ├── (dev) libdd-networking v1.0.0 (*)
      ├── libdd-remote-config v8.0.0 (*)
      ├── (dev) libdd-trace-protobuf v5.1.1
      │   ├── libdd-ffe v2.0.0 (*)
      │   └── libdd-remote-config v8.0.0 (*)
      ├── tokio-rustls v0.26.0
      │   ├── hyper-rustls v0.27.7 (*)
      │   └── (dev) libdd-networking v1.0.0 (*)
      ├── tokio-util v0.7.12
      │   ├── h2 v0.4.19 (*)
      │   ├── libdd-ffe v2.0.0 (*)
      │   └── libdd-remote-config v8.0.0 (*)
      └── (dev) tools v49.0.0
          └── (dev) libdd-networking v1.0.0 (*)

error[unmaintained]: chrono is unmaintained
   ┌─ /home/runner/work/libdatadog/libdatadog/Cargo.lock:36:1
   │
36 │ chrono 0.4.38 registry+https://github.com/rust-lang/crates.io-index
   │ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unmaintained advisory detected
   │
   ├ ID: RUSTSEC-2026-0543
   ├ Advisory: https://rustsec.org/advisories/RUSTSEC-2026-0543
   ├ The chrono crate is unmaintained; all versions are affected.
     
     Recommended alternatives:
     
     - [jiff](https://crates.io/crates/jiff)
     - [time](https://crates.io/crates/time)
   ├ Announcement: https://github.com/chronotope/chrono/issues/1768
   ├ Solution: No safe upgrade is available!
   ├ chrono v0.4.38
     └── libdd-ffe v2.0.0

advisories FAILED, bans ok, sources ok

Updated: 2026-11-18 02:25:46 UTC | Commit: bc917b3 | dependency-check job results

@datadog-datadog-prod-us1

datadog-datadog-prod-us1 Bot commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

Pipelines  Tests

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

🎯 Code Coverage (details)
• Patch Coverage: 87.50%
• Overall Coverage: 82.80%

This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 478310f | Docs | View more details | Give us feedback!

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant