Files
PhantomPtrandClaude Opus 5.5 9e5c258ad8 6.0.0: CI fixes, upgrade clean-up, docs and version (#377)
* docs: 6.0.0 changelog, README, FAQ and contributor docs

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* Release prep: version 6.0.0

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(ci): keep the leaked test pools reachable for LSan; skip the tiny-file floor under coverage

- ava1-ctest diag/mgmt_fs: the &'static Pool each test leaks for
  AvaTransport::with_pool is now held in a static list, so LeakSanitizer sees
  it (and the sessions on its threads) as reachable instead of reporting
  ~140 KB of indirect leaks the frame-name suppression could never match.
  Verified under clang ASan+LSan in Linux Docker: diag 13/13, mgmt_fs 40/40.
- download.rs tiny-file throughput floor is skipped when cargo-llvm-cov runs
  (CARGO_LLVM_COV set): instrumentation slows every call several times over.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): data_rust servers use the clients' fast timing

Five folder-host tests gave the client fast timing (dead after 500 ms) but
left the server on the default (ping every 2 s). On a slow runner a half-second
gap made the client drop healthy lanes and the job stalled until the 36 s
progress limit (both CI runs on #374/#377). The ordered-download tests already
set both ends; these now do too. The app uses one timing on both ends.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): the C lane socket-buffer test accepts Linux's wmem_max cap

Same cap as the Rust lane test: Linux CI gives the send side 416 KiB, so the
1 MiB floor failed under ASan in CI. 128 KiB is still far above a default.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(ci): gcc unused-result in the FTP selftest, serial ctest in make test, sender state on failure

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(ava1): flush a partial bundle after any reader message, not only a record

The last small-file record could arrive while a large file's chunk or root
was still queued behind it: the flush saw a non-empty channel, the next
message was not a record, and the records sat unsent until the receiver
cut the job as stalled (CI-only flake; 7/36 under local parallel load, 0/48 after).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(validate): run the C interop tests serially in make ci too

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(ava1): a resumed file goes in one-group pieces so a flapping link still makes progress

With kills every 2 s, the 4 MiB tail piece was cut before it was durable on every
attempt and the upload retried the same 28 MiB mark until it timed out (1 in 3
locally, CI coverage). Resumed files now send 1 MiB pieces: 5/5 in ~6 s.

Also: make docker-engine uses the repo root as context (as CI); governor
classification asserts are skipped under the sanitizers, where the C is the bottleneck.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): single-instance gate tests hold the C-server lock; relay test cap leaves margin

Coverage runs the C tests in parallel; the two gate tests stop the one global C server.
The relay order test's 4 MiB/s cap finished in 5.9 s on CI against its >6 s precondition.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): periodic-kill bound widened under coverage; tiny-file floor relaxed on CI runners

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): lane-kill test kills until a resend; periodic-kill liveness race skipped under coverage

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): a hung relay fails its test at the bound instead of hanging the CI job

The dest-ends relay test ran the relay under spawn_blocking: on a hang the timeout
panicked but the runtime then waited forever for the blocking thread (27 min on CI).
It now runs on a plain thread. The engine test steps get a 25-minute bound.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): dest-ends relay test in its own binary on small files; relay logs its silent retries

The test hung a CI job twice inside the shared relay binary (behind the 384 MiB relays'
lock) and never reproduced locally or in Linux Docker. It now runs alone, on 4x32 MiB
(still twice the relay buffer), ~3 s. The relay's open-dropped and retry paths now say
so on stderr instead of looping silently. Governor classification asserts also skip
under coverage, where the instrumented sender is the bottleneck.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): the partly-durable relay test caps its first attempt so it can be interrupted

On a fast runner the uncapped relay finished all 160 MiB before the first durable
report was observed. The first attempt now reads A through a 32 MiB/s proxy (~5 s);
the resume runs uncapped.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* feat: one-time 6.0 upgrade clean-up that keeps every setting

Host (engine, once per data dir): pre-5.x files no 5.41/6.0 code reads (api_token,
ps5upload.ini, ps5upload_profiles.ini/.json, app-config.json) move to ~/.ps5upload/legacy-5x/.
App (desktop, once): the 1.x/2.x config.json/profiles.json move to legacy-5x/; the cached
ezremote-dpi.elf nothing loads is deleted. Client: the retired upload-streams key is dropped.
Console (once per console, on its first 6.0 helper answer): the FTX2 /data/ps5upload/tx and
/spool folders are deleted; a failure retries later. The health check no longer lists those
folders (its fix would have recreated them). Nothing current is moved or reset.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): the download periodic-kill race is skipped under coverage, like its upload twin

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): the tiny-zip floor only catches a stall on CI runners

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(test): 'other calls still answer during launches' is a state check, not a wall-clock bound

Both mgmt_t6 tests asserted a sub-300 ms answer while a slow Sony call ran; on a loaded
runner that separates nothing. They now assert the answer arrived while the slow call
was still in flight.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

* fix(ci): reword the 6.0 migration comments for check-no-ftx2

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XR8fQMhp7Rdgq2e4U9oDff

---------

Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-05 07:04:11 -07:00

6.1 KiB

Engine

The Rust workspace: everything between the desktop app and the PS5.

The engine is a local HTTP service. The Tauri app doesn't talk to the console directly — it calls the engine, which speaks AVA1 to the payload. That split is deliberate: the same API is reachable from a browser, a script, or CI, so nothing the app can do is locked inside the app. See the self-hosted-engine entries in ../FAQ.md.

client (Tauri/React)  ──HTTP──▶  ps5upload-engine :19113
                                      │ AVA1 (Noise XX, sealed frames)
                                      ▼
                            payload  :9120

Crates

Crate What it is
ava1 The AVA1 protocol: frames, generated message codecs, Noise handshake, sessions, lanes, pairing. Spec: protocol/ava1/SPEC.md.
ava1-gen Generates the Rust and C codecs from protocol/ava1/schema/ava1.toml. Run it after editing the schema.
ava1-ctest Builds the payload's AVA1 C on the host so cargo test checks C against Rust.
ava1-chaos A misbehaving TCP proxy (latency, bandwidth caps, blackholes, kills) for resilience tests.
ps5upload-core The bulk of the logic — socket helpers, the management seam, archive inspection and plan previews, filesystem and app RPCs, volume parsing, package install, saves, cheats, hardware, SMB, BPS patching.
ps5upload-engine The Axum HTTP service (~100 routes), job tracking, SSE progress events, and the desktop + mobile entry points.
ps5upload-pkg .pkg parsing — headers, entries, split-file sets.
ps5upload-lab CLI for driving the console by hand over AVA1: management calls, uploads, benchmarks (bench), pairing and chaos tools. Useful when you want one call, not a workflow.
ps5upload-tests Integration tests that run against in-process and loopback AVA1 peers, with no console.

Working on it

cargo test --workspace          # no PS5 required — loopback peers
cargo test -p ava1-ctest -- --test-threads=1   # payload AVA1 C vs Rust, on the host
cargo fmt --all -- --check      # CI fails on this; client/src-tauri has its own
cargo build --release -p ps5upload-engine

Run it standalone and point it at a console:

PS5_ADDR=<ip> cargo run -p ps5upload-engine
curl "http://127.0.0.1:19113/api/ps5/status?addr=<ip>"

PS5UPLOAD_ALLOW_IP lets another machine reach it. The API is unauthenticated and can read, write and delete files on the console — keep it on a trusted LAN.

Testing

engine/crates/ps5upload-tests/tests/ runs against a loopback AVA1 console (the Rust job host plus a scripted management node): single-file, folder, small-file, resume-after-drop, retry classification, exclude rules, .zip/.7z/.rar streaming, hardware commands, and volume parsing. Unit tests live beside their modules.

Anything touching real transfer behaviour still needs hardware — see ../TESTING.md.

Things that will bite you

  • The payload emits JSON; serde parses it. Field names must be snake_case or serde silently leaves the field at its default — a zero or an empty string, not an error. One malformed byte rejects the whole response, so anything interpolated into a payload-side JSON string has to be escaped first.
  • ok: false inside a 200 is a refusal, not a success. Several payload commands answer that way. client/src/api/ps5.ts has an assertOk() for action endpoints; status endpoints keep ok:false as data because there it means "unsupported on this console".
  • One port, one session. Everything goes to 9120 as AVA1: management calls on the session's control connection, bulk data on its lanes.
  • A stream install has the PS5 fetch the package from the engine. The engine tells the console an address to connect to; behind Docker's bridge network, a VPN or a virtual adapter that address is one the PS5 cannot reach. PS5UPLOAD_PKG_HOST_IP pins the LAN IP the console is given (publish port 19113 when the engine is in a container; compose.yaml uses host networking instead). Every stream-unreachable failure — the pre-install reach check, a Sony refusal before any byte was fetched, and an accepted install the console never fetched from — ends in install/mod.rs stream_unreachable_hint_for, which names this variable.

Where the web UI's file browser opens (PS5UPLOAD_BROWSE_ROOTS)

The browser-based UI picks local files and folders from the engine's filesystem. By default that picker opens at the engine's home directory, which in the Docker images is /data, the engine's own state volume, not the folder where you mounted your packages. Set PS5UPLOAD_BROWSE_ROOTS to a comma-separated list of directories (for example /pkgs, or /pkgs,/games) and the picker starts there instead. compose.yaml sets it to /pkgs, next to the volume mount it describes.

  • Unset or blank keeps the old behaviour (the home directory).
  • An entry that is not a directory the engine can see is skipped and named in the engine log (a warning), so a typo or a volume that was never mounted is visible. If no entry is usable the picker falls back to the home directory.
  • It only chooses where the picker opens. It is not a sandbox: the engine can still be asked for any path it can read.
  • HOME is deliberately left alone, because FPKG conversions write under $HOME/Downloads/fpkgs and a package mount is usually read-only.

Synology and other NAS boxes (saved servers: "Permission denied")

The Docker images keep their state (saved servers, install history, artwork cache) in /data and run as UID:GID 65532 unless you override user:. If you bind-mount a host folder that someone else owns, writing fails and the engine says so, naming the folder, the UID:GID it runs as, and the owner it found (#361). Fix it one of two ways:

  • set user: in the compose file to the folder's owner (ls -ln on the host shows the numbers), for example user: "1234:77777"; or
  • chown -R 65532:65532 <host folder> for the mount.

USER_ID / GROUP_ID environment variables are not read by these images; use user:. The account must also be allowed to write that folder in the NAS's own permission settings.