Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
21 commits
Select commit Hold shift + click to select a range
fb35dd4
README: add Also for LG webOS section linking Clean Your TV (#1369)
GLinnik21 Nov 13, 2026
4c00cf1
Crash reports: a record left by an earlier build is not filed under t…
GLinnik21 Nov 13, 2026
2d565dc
URI on webOS 5: admit firmware major 5 to the exported-window table (…
GLinnik21 Nov 13, 2026
4abce8d
Preview: a trailer halted before its landing finishes its request Can…
GLinnik21 Nov 13, 2026
bc8bafc
Troubleshooting: slow or deep audio on a soundbar means Digital Sound…
GLinnik21 Nov 13, 2026
ed3c720
Preview: a refused trailer's report names which condition refused it …
GLinnik21 Nov 13, 2026
fb3b3ff
Freeze reports name the main loop's leg, and a stall with no app fram…
GLinnik21 Nov 13, 2026
fbec5ed
webOS version: read /usr/lib/os-release when nyx's os_info.json is un…
GLinnik21 Nov 13, 2026
8e0f66a
Distribution: LG's seller guide gates native apps on a contract, and …
GLinnik21 Nov 13, 2026
6d7b48e
Deps: bump ed25519-compact to 2.6.1 (hold zip at 9.2.2)
dependabot[bot] Nov 13, 2026
6f9bbc9
Deps: bump the Rust toolchain from 1.99.1 to 1.100.0
dependabot[bot] Nov 13, 2026
6a74d2a
refactor(ui): complete owned input and frame scheduling
GLinnik21 Nov 13, 2026
5ca2777
Net: bound a never-pausing peer's cancel latency by what the transpor…
GLinnik21 Nov 14, 2026
a3ea02c
Media: the window-reread test times its wait from before the stop's s…
GLinnik21 Nov 14, 2026
a680294
Harness: cheaper wall gate, fewer DepGates runs, slowest python gates…
GLinnik21 Nov 14, 2026
2b76740
Tests: the 8 slowest unit tests do the same checks for a fraction of …
GLinnik21 Nov 14, 2026
a8e460b
CI: build the host crates with line tables only (#1383)
GLinnik21 Nov 14, 2026
f5bb31d
Check: run the hostsim unit pass only over the crates that declare ho…
GLinnik21 Nov 14, 2026
2e1b78d
Harness: each DepGates self-test runs only the rule it plants for (#1…
GLinnik21 Nov 14, 2026
0a6e505
Merge remote-tracking branch 'upstream/main' into upstream-merge-2026…
tiper Nov 14, 2026
c66a259
Jellyfin: favourites wake the frame through the worker door
tiper Nov 14, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions .agents/skills/which-tier/SKILL.md
Original file line number Diff line number Diff line change
Expand Up @@ -100,9 +100,11 @@ cargo can see a python file).
everybody quotes is `cargo test --lib` alone, and that part is itself tens of seconds now (the per-crate test binaries, `UNIT_SUITE` in the Makefile, run side by side through
`tools/cargo-test-parallel.py`; `make test-crate C=<crate>` runs just the one you touched); end to end the
gate runs in MINUTES warm. The bulk of it is `python3 tests/test_harness.py`, which shells out to
`ci/check-deps.sh` (11 s a run) once or more per self-test to prove each structure gate still
catches a planted violation — 661 s of a 1100 s run, measured 2026-10-01 on a cold lane at load
~5, before those self-tests moved to private tree copies and a thread pool (175 s after). The lab selftest's **~7 s is mostly two DELIBERATE rate-limit waits** — so
`ci/check-deps.sh` to prove each structure gate still catches a planted violation: each DepGates
self-test runs only the rule or rules it plants for (`--only <rules>`), and only the whole-script
tests, with the one shared run over the real tree, run every rule. That was 661 s of a 1100 s run,
measured 2026-10-01 on a cold lane at load ~5, before those self-tests moved to private tree
copies and a thread pool (175 s after). The lab selftest's **~7 s is mostly two DELIBERATE rate-limit waits** — so
a ten-second-plus pause there is the target working, not a hang. (That step
also SSDP-probes the LAN to report whether a UPnP gateway is present; still no television, still no
lock.) Do not write a new number here: measure it if you need one — this file has already carried a
Expand Down
31 changes: 31 additions & 0 deletions .github/actions/host-rust/action.yml
Original file line number Diff line number Diff line change
Expand Up @@ -42,8 +42,39 @@ runs:
# is byte-for-byte the one `make check` ran against, and the ARM cross-build's.
uses: ./.github/actions/rust-toolchain

- name: Build the host crates with line tables only
# Nothing in CI attaches a debugger to a host test binary, and a panic backtrace needs only the
# line table. The default `dev`/`test` profile writes full DWARF for every crate, dependencies
# included, which is most of what rustc's codegen, the linker and the cache archive carry for
# the 18 test binaries. This is the per-user cargo config, so it reaches the bare `cargo`
# lines and every `make` target of the host jobs alike. A dependency built with the old
# profile is a different unit to cargo, and rust-cache's key does not see this file (it hashes
# the manifests, the lockfile, the toolchain files and the workspace's `.cargo/config.toml`,
# not the home one), so the cache step below carries its own `prefix-key`: without it the
# full-DWARF cache restores as an exact hit, every dependency is rebuilt, and the post step
# finds the cache up to date and saves nothing. It is NOT the manifest: a developer's checkout
# keeps full DWARF (docs/agent-reference.md, the lane policy), and the profiles the ARM binary
# is built with do not inherit `dev`, so the shipped library is untouched.
shell: bash
run: |
mkdir -p "$HOME/.cargo"
cat >> "$HOME/.cargo/config.toml" <<'EOF'
[profile.dev]
debug = "line-tables-only"
[profile.dev.package."*"]
debug = false
EOF

- uses: Swatinem/rust-cache@d1527c8f67ef9af357290475d4af4b0fab8f12e0 # v2
# The cache key already carries the calling job's id, so the three Rust jobs each keep their
# own `target/` instead of racing to save one.
#
# The `prefix-key` names the profile the cached `target/` was built with: the line-tables
# config above is not an input to rust-cache's key, so the profile has to be in the prefix.
# Change that config and bump the number here, or the old entry restores as an exact hit. The
# entry is built cold once per lockfile or toolchain change, like any other key of this cache.
# The host jobs' old `v0-rust-` entries are never read again; GitHub drops an entry that
# nothing has restored for seven days.
with:
prefix-key: v1-rust-linetables
workspaces: rust-modules
2 changes: 1 addition & 1 deletion .github/workflows/ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -454,7 +454,7 @@ jobs:
name: host harness tests
needs: changes
if: needs.changes.outputs.gated == 'true'
runs-on: ubuntu-24.04
runs-on: ubuntu-latest
# FinNative: 15, not 10. Its larger tree takes check-deps' plants 8-10 min on a runner.
timeout-minutes: 15
steps:
Expand Down
16 changes: 14 additions & 2 deletions Makefile
Original file line number Diff line number Diff line change
Expand Up @@ -1692,6 +1692,18 @@ check-cargo-lint: lint
# ci/test_test_crate.py pins the two equal.
UNIT_RUN = python3 ../tools/cargo-test-parallel.py
UNIT_SUITE = plxnative-modules plx_base plx_machine plx_platform plx_gfx plx_net plx_ui plx_plex plx_telemetry plx_data plx_session plx_media plx_playback_contract plx_play_uri plx_play_feed plx_media_codecs plx_appkit plx_screens
# The `hostsim` pass runs only the crates that DECLARE a `hostsim` feature (HOSTSIM_SUITE, a subset of
# UNIT_SUITE in the same order; ci/test_test_crate.py pins both the subset and the recipe). The seven that
# do not (plx_net, plx_ui, plx_appkit, plx_screens, plx_media_codecs, plx_play_uri, plx_playback_contract)
# have no `feature = "hostsim"` in their source and the same test names in both passes (14 crates did in
# CI run 43528579834; the three that differ are `plxnative-modules`, `plx_media` and `plx_play_feed`,
# all kept). Their own test set is therefore identical in the two passes. The old second run did link
# their test binaries against hostsim-built `plx_base`, `plx_gfx`, `plx_platform`, `plx_media` and
# `plx_play_feed` (steerable `ENV_STEERABLE` paths, larger glyph-cache sizes, `ffi_host` present), but
# none of their tests observes those dependency-side differences; the guard does not check that, it
# checks that the crate declares no `hostsim` feature and has no `feature = "hostsim"` in its src, and
# fails when one gains either, which is the moment it must join the list.
HOSTSIM_SUITE = plxnative-modules plx_base plx_machine plx_platform plx_gfx plx_plex plx_telemetry plx_data plx_session plx_media plx_play_feed

# The default-feature unit suite and everything that drives cargo through ci/ self-tests.
check-cargo-unit-default:
Expand Down Expand Up @@ -1743,7 +1755,7 @@ check-cargo-unit-default:

# The `hostsim` unit suite (the host feed seam exists only there).
check-cargo-unit-hostsim:
@# The SAME suite as `check-cargo-unit-default`'s `cargo test --lib`, again under `hostsim`, which is
@# The crates of HOSTSIM_SUITE (the ones that declare `hostsim`) again under `hostsim`, which is
@# not a duplicate run: the host feed seam (`play-feed/src/ffi_host.rs`) only exists in that configuration,
@# so every test that drives an AU through `sf_feed` is COMPILED OUT of the default pass and cannot fail it. The prime-livelock
@# regression is one of those, and it guards the 94-second freeze of 2026-08-29 — a defect that
Expand All @@ -1754,7 +1766,7 @@ check-cargo-unit-hostsim:
cd rust-modules && PATH="$$HOME/.cargo/bin:$$PATH" PLXNATIVE_RUNTIME_DIR="$$d" \
$(TELEMETRY_ENV) \
$(HOST_ENV) \
$(UNIT_RUN) -- cargo +$(RUST_TOOLCHAIN) test --lib -p plxnative-modules -p plx_base -p plx_machine -p plx_platform -p plx_gfx -p plx_net -p plx_ui -p plx_plex -p plx_telemetry -p plx_data -p plx_session -p plx_media -p plx_playback_contract -p plx_play_uri -p plx_play_feed -p plx_media_codecs -p plx_appkit -p plx_screens --features hostsim
$(UNIT_RUN) -- cargo +$(RUST_TOOLCHAIN) test --lib -p plxnative-modules -p plx_base -p plx_machine -p plx_platform -p plx_gfx -p plx_plex -p plx_telemetry -p plx_data -p plx_session -p plx_media -p plx_play_feed --features hostsim

# OPT-IN incremental inner loop: the default-feature unit suite (the same `cargo test --lib` as
# `check-cargo-unit-default`, same throwaway runtime root and telemetry env) with
Expand Down
31 changes: 25 additions & 6 deletions PRIVACY.md
Original file line number Diff line number Diff line change
Expand Up @@ -201,10 +201,19 @@ or `development`), the time of the panic (read and bounded as above), the same k
webOS and hardware compatibility details (including the `probe` fact and the firmware `build`) that the fault fallback report carries, and
the Crash report ID below. It carries none of the library details above.

A main-thread freeze (the app not responding for five seconds or more while it is open on screen) is reported like a crash, under the same choice: code addresses and register values, one of three duration ranges (5 to 10 s, 10 to 30 s, 30 s or more) or "unrecovered", and the details above, without stopping the app.
A main-thread freeze (the app not responding for five seconds or more while it is open on screen) is reported like a crash, under the same choice: code addresses and register values, one of three duration ranges (5 to 10 s, 10 to 30 s, 30 s or more) or "unrecovered", the step of the app's own frame loop the app was in when it stopped responding (`freeze.leg`: one fixed word from a list of eleven naming a phase of the loop, never anything about what was on screen or playing: `event-ingest`, `results`, `dev-scenarios`, `playback-pump`, `land-results`, `navcommit`, `dispatcher-frame`, `update`, `prepare-window`, `present-and-swap` or `frame-report`), and the details above, without stopping the app.

When another part of the television's software ends PlxNative by calling `exit()` (for example the system closing it), the app records the call and reports it like a crash, under the same choice: the exit status (a number), which thread made the call (its number, whether it was the main thread, and its own name, a technical label set by PlxNative or a library), the library the call came from (its file name, never the folder, with its build identifier, load address and size, as for a crash report above) and the code address of the call, the time of the call to the second, and the details above.

A report built from the crash log (a fault, a Rust panic, a freeze or an `exit()` call, as above) is
sent the next time the app starts, which after an update is a newer build than the one that left
the record. When the record names a different build of PlxNative than the one sending it, the report's
`release` (the app version it is filed under) is left out rather than filed under the newer
version, and it carries two tags instead: `crash.uploaded_by`, the app version that sent it, and
`crash.build_id`, the build identifier of the build that crashed. Both are an app version and a
binary build identifier, which these reports already carried; nothing new is collected. A report from the same
build as the one sending it, or one whose build cannot be compared, carries `release` and neither tag.

Every crash and error report carries a **Crash report ID**: a random identifier created on this
television when you turn crash reports on, sent as the report's `user.id`. It exists so that
repeated crashes under one Crash report ID are counted once rather than once each — Sentry's
Expand Down Expand Up @@ -287,11 +296,21 @@ after the app returned to the foreground) / `preview` (the trailer that plays qu
detail page, which the viewer did not press Play for); and `attempt_ordinal`, which open of this
playback it was, `1` to `9` and then `9+`, counting the first open, every retry and every handoff.
The count starts again at `1` with each new Play and is kept only in memory. Both appear as a tag
and in the playback context. A trailer preview that fails is reported with the same fields and no
more: it is marked `preview`, it counts as the first open, it names direct play as the attempted
delivery (the only route a preview ever opens), and its `started` and `attempt_started` fields
say whether the trailer actually showed a picture before it failed. Sentry files it in an issue of
its own, apart from playback failures.
and in the playback context. A trailer preview that fails is reported with the same fields and one
more, below: it is marked `preview`, it counts as the first open, it names direct play as the
attempted delivery (the only route a preview ever opens), and its `started` and `attempt_started`
fields say whether the trailer actually showed a picture before it failed. Sentry files it in an
issue of its own, apart from playback failures.

A trailer preview that the app refuses to open, because a preview only ever plays an unmodified
direct play (the failure kind `preview_not_direct`), also names which condition refused it, as one
closed tag, `preview_refusal`: `no_part` (the item has no media file to play), `not_direct` (a direct
play was not available, so the route would have been a transcode or a remux, whatever the reason:
the server's answer, the connection's or the quality setting's rules, or a track the television
cannot play as it is) or `adaptive` (a direct play was available, but Auto quality chose a
bandwidth-adapted stream instead). The first condition that fails in that order is the one named.
No other failure carries it, and it appears as a tag only, not in the playback context. It is one
of those three words and nothing the server said.

The report also names `backend` — `unknown` / `feed` / `uri` — and `terminal_stage` —
`unknown` / `planning` / `open` / `decode` / `presentation` / `release`. These describe only the
Expand Down
6 changes: 3 additions & 3 deletions THIRD-PARTY-NOTICES.md
Original file line number Diff line number Diff line change
Expand Up @@ -314,7 +314,7 @@ stated; election does not alter your rights under the other arms.

| Package | Version(s) | Declared licence | Copyright |
|---|---|---|---|
| The Rust standard library (`core`, `alloc`, `std`, `panic_unwind`, `unwind`, and the `rustc-std-workspace-*` shims) | rustc 1.99.1 (58cf8f9c3 2026-10-13) | MIT OR Apache-2.0 | The Rust Project Developers |
| The Rust standard library (`core`, `alloc`, `std`, `panic_unwind`, `unwind`, and the `rustc-std-workspace-*` shims) | rustc 1.100.0 (0ef1bc386 2026-11-09) | MIT OR Apache-2.0 | The Rust Project Developers |
| — its in-tree backtrace support (`library/backtrace`) | 0.3.76 | MIT OR Apache-2.0 | 2014 Alex Crichton |
| — its in-tree mpmc channels (`library/std/src/sync/mpmc`) | in-tree | MIT OR Apache-2.0 | 2019 The Crossbeam Project Developers |
| addr2line | 0.27.1 | Apache-2.0 OR MIT | The `addr2line` authors |
Expand All @@ -329,7 +329,7 @@ stated; election does not alter your rights under the other arms.
| crc | 3.4.0 | MIT OR Apache-2.0 | 2017 crc-rs Developers |
| crc-catalog | 2.5.0 | MIT OR Apache-2.0 | Akhil Velagapudi |
| crc32fast | 1.5.0 | MIT OR Apache-2.0 | 2018 Sam Rijs, Alex Crichton and contributors |
| ed25519-compact | 2.6.0 | MIT | 2020-2026 Frank Denis |
| ed25519-compact | 2.6.1 | MIT | 2020-2026 Frank Denis |
| either | 1.18.0 | MIT OR Apache-2.0 | Copyright (c) 2015 (the upstream notice names no holder) |
| encoding_rs | 0.8.44 | (Apache-2.0 OR MIT) AND BSD-3-Clause | Mozilla Foundation. Its encoding tables derive from the WHATWG Encoding Standard, whose data files are BSD-3-Clause (`licenses/WHATWG-BSD-3-Clause.txt`) |
| equivalent | 1.0.3 | Apache-2.0 OR MIT | 2016–2023 The equivalent authors |
Expand Down Expand Up @@ -381,7 +381,7 @@ stated; election does not alter your rights under the other arms.
| tap | 1.0.1 | Copyright (c) 2017 Elliot Linder |
| wyz | 0.5.1 | Copyright (c) 2018 myrrlyn (Alexander Payne) |
| zmij | 1.0.21 | David Tolnay (a Rust port of Victor Zverovich's C++ `zmij`) |
| rust-lang/libm — compiled *inside* `compiler_builtins` and exposed as intrinsics on this target | as vendored in rustc 1.99.1 | (c) 2018 Jorge Aparicio; musl libc (c) 2005-2020 Rich Felker, et al.; CORE-MATH; and the fdlibm-derived notice: (c) 1993, 2004 Sun Microsystems; (c) 2003-2011 David Schultz; (c) 2003-2009 Steven G. Kargl; (c) 2003-2009 Bruce D. Evans; (c) 2008 Stephen L. Moshier; (c) 2017-2018 Arm Limited |
| rust-lang/libm — compiled *inside* `compiler_builtins` and exposed as intrinsics on this target | as vendored in rustc 1.100.0 | (c) 2018 Jorge Aparicio; musl libc (c) 2005-2020 Rich Felker, et al.; CORE-MATH; and the fdlibm-derived notice: (c) 1993, 2004 Sun Microsystems; (c) 2003-2011 David Schultz; (c) 2003-2009 Steven G. Kargl; (c) 2003-2009 Bruce D. Evans; (c) 2008 Stephen L. Moshier; (c) 2017-2018 Arm Limited |

<!-- END GENERATED notices:mit-only -->

Expand Down
Loading
Loading