chore(release): bump workspace version to 0.9.1

Release 0.9.1 — a patch release led by three field-reported crash/brick
fixes, a security-review hardening pass, and the tiered stats overlay
landing on every platform.

The regression that forces the release: 0.9.0's stats-HUD display stage
hard-linked an API-33 NDK symbol believed API-26, so the native library
failed to load on every Android < 13 device — "Identity unavailable:
…NativeBridge", dead pair button, no discovery. The callback is now
dlsym-resolved (pre-13 devices simply lose that one HUD row), and a new CI
guard fails the APK build if the native lib ever imports a symbol beyond
the API-28 floor.

Two more crash-class host fixes ride along. GNOME: one console Approve
admitted every parked knock a retrying client had stacked up — three
instant Mutter virtual monitors segfaulted gnome-shell down to the GDM
greeter; knock generations now admit exactly one, Mutter topology
mutations are serialized process-wide, stale session-env is cleared, and
an opt-in PUNKTFUNK_RECOVER_SESSION_CMD hook can revive a dead session on
the next connect. Windows: twin identical GPUs died in a TEX_FAIL retry
loop when the IDD-push ring opened on the wrong twin — the ring now
rebinds once onto the render LUID the driver reports, and GPU inventory
ordering is LUID-sorted and stable for the boot.

Input: held gamepad state is rock-solid — Android pins one qualified pad
device (motion-sensor siblings zeroed axes on every event interleave), and
pad state now travels as sequence-gated idempotent snapshots
(HOST_CAP_GAMEPAD_STATE, negotiated; either end older falls back to the
per-transition path).

Security review findings addressed: sliding-window anti-replay on the
AEAD-authenticated data-plane sequence, 0600 client key + 0700 config dir
(existing stores re-locked), the web console's post-login open redirect
closed, the Windows installer's FFmpeg DLLs SHA-256-pinned, and a fork
guard on PR jobs sharing signing-runner labels.

Features: the 3-tier stats overlay (Compact/Normal/Detailed) ships on
every platform with cycle shortcuts and settings pickers; Apple gains the
experimental stage-3 glass-gated presenter (fixes the hidden 29-30 ms
display-stage queue on 120 Hz ProMotion; stage 2 stays default) and a
morphing, device-corner-concentric overlay card. Plus repo-wide docs sync
and the flatpak CI network hardening.

The [workspace.package] version (inherited by every crate via
version.workspace) is the release being cut; the 14 workspace entries in
Cargo.lock and the api/openapi.json info.version are refreshed to match
(CI builds --locked). Canary derives from the tag as minor+1 of the latest
stable.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

.gitea/workflows/apple.yml

@@ -16,8 +16,17 @@ on:
   workflow_dispatch:
 
 jobs:
+  # SECURITY: builds/tests PULL-REQUEST code on the host-mode, persistent `macos-arm64` runner shared
+  # with the release-signing job (release.yml, which loads the App Store Connect key). Untrusted PR
+  # code could persist on it or harvest signing material. Definitive fix is server-side: enable Gitea's
+  # "require approval for PRs from outside collaborators/forks", and/or isolate PR CI on ephemeral
+  # runners. The `if:` is a fail-open backstop — it skips fork PRs where Gitea reports the fork flag and
+  # still runs same-repo PRs (and where the flag is absent), so it never blocks internal PR CI.
   swift:
     runs-on: macos-arm64
+    if: >-
+      gitea.event_name != 'pull_request' ||
+      gitea.event.pull_request.head.repo.fork != true
     timeout-minutes: 60
     steps:
       - uses: actions/checkout@v4

.gitea/workflows/flatpak.yml

@@ -73,7 +73,10 @@ jobs:
           dnf -y install flatpak flatpak-builder git python3 python3-aiohttp python3-tomlkit curl jq \
             gnupg2 rsync openssh-clients
           # Flathub provides the GNOME runtime/SDK + the rust-stable + ffmpeg-full extensions.
-          flatpak remote-add --user --if-not-exists flathub \
+          # retry.sh: the busy runner intermittently drops DNS lookups ("[6] Could not resolve
+          # hostname" seconds after dnf pulled 300+ packages fine) — never fail the job on a
+          # single-shot fetch. Same treatment for every network command below.
+          bash scripts/ci/retry.sh 5 flatpak remote-add --user --if-not-exists flathub \
             https://dl.flathub.org/repo/flathub.flatpakrepo
           git config --global --add safe.directory "$PWD"
 
@@ -108,7 +111,7 @@ jobs:
         # device" (see packaging/flatpak/prune-windows-lock.py). The committed Cargo.lock is
         # untouched; cargo --offline only needs sources for the crates it compiles.
         run: |
-          curl -fsSL -o /tmp/flatpak-cargo-generator.py \
+          curl -fsSL --retry 5 --retry-all-errors --retry-delay 5 -o /tmp/flatpak-cargo-generator.py \
             https://raw.githubusercontent.com/flatpak/flatpak-builder-tools/master/cargo/flatpak-cargo-generator.py
           python3 packaging/flatpak/prune-windows-lock.py Cargo.lock /tmp/Cargo.flatpak.lock
           python3 /tmp/flatpak-cargo-generator.py /tmp/Cargo.flatpak.lock \
@@ -142,18 +145,43 @@ jobs:
           DEST="${DEPLOY_USER}@${DEPLOY_HOST}"
           mkdir -p "$PWD/repo"
           # Pull the currently-published repo (all channels' objects + refs) into the repo the build
-          # will extend. No --delete: the local repo starts empty, so this only ADDS. A missing
-          # server repo (very first publish) is fine — we continue with a fresh repo.
-          rsync -az --info=stats1 -e "$SSH" "$DEST:$DEPLOY_DIR/site/repo/" "$PWD/repo/" \
-            || echo "::warning::no published repo to seed (first publish?) — continuing fresh"
+          # will extend. No --delete: the local repo starts empty, so this only ADDS.
+          # Probe first (retried) whether a published repo exists at all: ONLY that case may
+          # continue with a fresh repo. A transient network failure must FAIL the job instead —
+          # a blanket `rsync || continue` here is exactly how a flaky link produces the
+          # single-branch summary that clobbers the other channel (the bug described above).
+          PRESENT=$(bash scripts/ci/retry.sh 5 $SSH "$DEST" \
+            "[ -d $DEPLOY_DIR/site/repo/refs ] && echo present || echo absent")
+          if [ "$PRESENT" = present ]; then
+            bash scripts/ci/retry.sh 5 rsync -az --info=stats1 -e "$SSH" \
+              "$DEST:$DEPLOY_DIR/site/repo/" "$PWD/repo/"
+          else
+            echo "::warning::no published repo on the server (first publish) — continuing fresh"
+          fi
           echo "seeded refs:"; ls "$PWD/repo/refs/heads/app/$APP_ID/x86_64/" 2>/dev/null || echo "  (none)"
 
-      - name: Build the flatpak (install deps from Flathub, offline build)
+      - name: Prefetch deps + sources (retried — the network phase, split off the build)
         run: |
-          # --install-deps-from=flathub pulls everything the manifest declares: the GNOME 50
-          # runtime/SDK + the rust-stable (//25.08, rustc 1.96) and llvm20 SDK extensions, plus
-          # the runtime's auto codecs-extra (HEVC libavcodec). --disable-rofiles-fuse is the
-          # container-safe path (no FUSE).
+          # All of the job's heavy network I/O happens HERE, retried, so a dropped DNS lookup
+          # or TCP dial costs a backoff-retry instead of the whole (long) compile:
+          #   1) --install-deps-only pulls everything the manifest declares from Flathub: the
+          #      GNOME 50 runtime/SDK + the rust-stable (//25.08, rustc 1.96) and llvm20 SDK
+          #      extensions, plus the runtime's auto codecs-extra (HEVC libavcodec).
+          #   2) --download-only fetches every source (all crates in cargo-sources.json) into
+          #      the .flatpak-builder state dir. Both are resumable/idempotent, so re-running
+          #      after a partial failure is safe and cheap.
+          # --disable-rofiles-fuse is the container-safe path (no FUSE).
+          bash scripts/ci/retry.sh 5 flatpak-builder --user --force-clean --disable-rofiles-fuse \
+            --install-deps-from=flathub --install-deps-only \
+            "$PWD/build-dir" "$MANIFEST"
+          bash scripts/ci/retry.sh 5 flatpak-builder --user --force-clean --disable-rofiles-fuse \
+            --download-only \
+            "$PWD/build-dir" "$MANIFEST"
+
+      - name: Build the flatpak (offline — deps + sources prefetched above)
+        run: |
+          # Everything is already local (state dir warmed by the prefetch step), so this long
+          # step needs no network; --install-deps-from stays as a no-op safety net.
           # --default-branch=$FLATPAK_BRANCH pins the ref to app/io.unom.Punktfunk/x86_64/<branch>
           # (canary or stable) so the matching hosted .flatpakref resolves deterministically
           # (manifest sets no branch).
@@ -272,11 +300,13 @@ jobs:
           printf '%s\n' "$DEPLOY_SSH_KEY" > ~/.ssh/deploy; chmod 600 ~/.ssh/deploy
           SSH="ssh -i $HOME/.ssh/deploy -p ${DEPLOY_PORT:-22} -o StrictHostKeyChecking=accept-new"
           DEST="${DEPLOY_USER}@${DEPLOY_HOST}"
-          $SSH "$DEST" "mkdir -p ~/$DEPLOY_DIR/site/repo"
-          rsync -az --info=stats1 -e "$SSH" repo/ "$DEST:$DEPLOY_DIR/site/repo/"
-          rsync -az -e "$SSH" site/unom.flatpakrepo "site/${APP_ID}.flatpakref" "site/${APP_ID}.Canary.flatpakref" site/index.html "$DEST:$DEPLOY_DIR/site/"
-          rsync -az -e "$SSH" packaging/flatpak/server/compose.production.yml packaging/flatpak/server/Caddyfile "$DEST:$DEPLOY_DIR/"
-          $SSH "$DEST" "cd ~/$DEPLOY_DIR && docker compose -f compose.production.yml up -d"
+          # All idempotent — retried because the runner's link to unom-1 drops TCP dials under
+          # load (the same flake that hits docker.yml's deploy-docs with "dial tcp: i/o timeout").
+          bash scripts/ci/retry.sh 5 $SSH "$DEST" "mkdir -p ~/$DEPLOY_DIR/site/repo"
+          bash scripts/ci/retry.sh 5 rsync -az --info=stats1 -e "$SSH" repo/ "$DEST:$DEPLOY_DIR/site/repo/"
+          bash scripts/ci/retry.sh 5 rsync -az -e "$SSH" site/unom.flatpakrepo "site/${APP_ID}.flatpakref" "site/${APP_ID}.Canary.flatpakref" site/index.html "$DEST:$DEPLOY_DIR/site/"
+          bash scripts/ci/retry.sh 5 rsync -az -e "$SSH" packaging/flatpak/server/compose.production.yml packaging/flatpak/server/Caddyfile "$DEST:$DEPLOY_DIR/"
+          bash scripts/ci/retry.sh 5 $SSH "$DEST" "cd ~/$DEPLOY_DIR && docker compose -f compose.production.yml up -d"
           echo "deployed → $REPO_URL/${APP_ID}.flatpakref"
 
       - name: Attach bundle to the Gitea release (stable tags only)

.gitea/workflows/windows-drivers.yml

@@ -30,8 +30,15 @@ on:
 # scripts/ci/ensure-windows-toolchain.ps1, a fast no-op once already present.
 
 jobs:
+  # SECURITY: builds PULL-REQUEST code on the host-mode, persistent `windows-amd64` runner shared with
+  # the release-signing jobs (windows-host.yml / windows-msix.yml). See windows.yml for the full
+  # rationale. Definitive fix is server-side (Gitea outside-collaborator approval + isolated PR
+  # runners); the `if:` is a fail-open backstop that never blocks internal PR CI.
   probe-and-proto:
     runs-on: windows-amd64
+    if: >-
+      github.event_name != 'pull_request' ||
+      github.event.pull_request.head.repo.fork != true
     timeout-minutes: 30
     defaults:
       run:
@@ -108,6 +115,9 @@ jobs:
   # DLL's FORCE_INTEGRITY (/INTEGRITYCHECK) bit — the M0 self-signed-load question.
   driver-build:
     runs-on: windows-amd64
+    if: >-
+      github.event_name != 'pull_request' ||
+      github.event.pull_request.head.repo.fork != true
     timeout-minutes: 45
     defaults:
       run:

.gitea/workflows/windows.yml

@@ -68,8 +68,20 @@ on:
   workflow_dispatch:
 
 jobs:
+  # SECURITY: this job builds PULL-REQUEST code (attacker-controllable build.rs / cargo build) on the
+  # host-mode, persistent `windows-amd64` runner that the release-SIGNING jobs (windows-host.yml /
+  # windows-msix.yml, which decrypt MSIX_CERT_PFX_B64 + REGISTRY_TOKEN to disk) also run on. Untrusted
+  # PR code could therefore persist on that machine or harvest signing material a later job exposes.
+  # The DEFINITIVE fix is operational and lives outside this file: enable Gitea's "require approval to
+  # run workflows for PRs from outside collaborators/forks", and/or route PR CI to isolated ephemeral
+  # runners. The `if:` below is only a backstop — it skips fork PRs where Gitea reports the fork flag,
+  # and FAILS OPEN (still runs) for same-repo PRs and on Gitea versions that don't populate it, so it
+  # never blocks internal PR CI.
   build:
     runs-on: windows-amd64
+    if: >-
+      github.event_name != 'pull_request' ||
+      github.event.pull_request.head.repo.fork != true
     timeout-minutes: 90
     strategy:
       fail-fast: false

CONTRIBUTING.md

@@ -40,4 +40,6 @@ Generated artifacts are checked in and CI fails on drift: `include/punktfunk_cor
 `api/openapi.json` (`cargo run -p punktfunk-host -- openapi`). Match the surrounding code's comment
 density and naming. Commit messages end with the `Co-Authored-By` trailer (see `git log`).
 
-See [`CLAUDE.md`](CLAUDE.md) for the full build/test/run guide and design invariants.
+See the [README's Build & test section](README.md#build--test-from-source) and
+[Design invariants](README.md#design-invariants) for the full build/test/run guide, and the
+[docs site](https://docs.punktfunk.unom.io) for architecture and per-platform guides.

Cargo.lock

@@ -2154,7 +2154,7 @@ dependencies = [
 
 [[package]]
 name = "latency-probe"
-version = "0.9.0"
+version = "0.9.1"
 
 [[package]]
 name = "lazy_static"
@@ -2286,7 +2286,7 @@ checksum = "0ceec5bc11778974d1bcb055b18002eba7f4b3518b6a0081b3af5f21666da9ad"
 
 [[package]]
 name = "loss-harness"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "punktfunk-core",
 ]
@@ -2765,7 +2765,7 @@ checksum = "9b4f627cb1b25917193a259e49bdad08f671f8d9708acfd5fe0a8c1455d87220"
 
 [[package]]
 name = "pf-client-core"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "async-channel",
@@ -2787,7 +2787,7 @@ dependencies = [
 
 [[package]]
 name = "pf-console-ui"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "ash",
@@ -2808,7 +2808,7 @@ dependencies = [
 
 [[package]]
 name = "pf-ffvk"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "ash",
  "bindgen",
@@ -2817,7 +2817,7 @@ dependencies = [
 
 [[package]]
 name = "pf-presenter"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "ash",
@@ -3001,7 +3001,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-client-android"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "android_logger",
  "jni",
@@ -3017,7 +3017,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-client-linux"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "async-channel",
@@ -3033,7 +3033,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-client-session"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "pf-client-core",
@@ -3048,7 +3048,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-client-windows"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "async-channel",
@@ -3072,7 +3072,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-core"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "aes-gcm",
  "bytes",
@@ -3103,7 +3103,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-host"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "aes",
  "aes-gcm",
@@ -3175,7 +3175,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-probe"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "mdns-sd",
@@ -3189,7 +3189,7 @@ dependencies = [
 
 [[package]]
 name = "punktfunk-tray"
-version = "0.9.0"
+version = "0.9.1"
 dependencies = [
  "anyhow",
  "ksni",

Cargo.toml

@@ -26,7 +26,8 @@ exclude = [
     "clients/android/native/vendor/ndk",
 ]
 
-# ndk 0.9.0 verbatim from crates.io plus ONE visibility change: `MediaCodec::as_ptr` made public
+# ndk 0.9.0 verbatim from crates.io plus ONE visibility change (and two warning fixes — an
+# unnecessary `std::` qualification and a feature-gated `Result` import): `MediaCodec::as_ptr` made public
 # (upstream keeps it private and exposes no frame-rendered binding), so the Android client can
 # call `AMediaCodec_setOnFrameRenderedCallback` via ndk-sys for the HUD's `display` stage
 # (design/stats-unification.md). Drop the patch when upstream exposes the pointer or the callback.
@@ -34,7 +35,7 @@ exclude = [
 ndk = { path = "clients/android/native/vendor/ndk" }
 
 [workspace.package]
-version = "0.9.0"
+version = "0.9.1"
 edition = "2021"
 rust-version = "1.82"
 license = "MIT OR Apache-2.0"

README.md

@@ -22,7 +22,8 @@ punktfunk pairs a **virtual-display streaming host** with native clients on ever
 the existing **GameStream** protocol, so any [Moonlight](https://moonlight-stream.org/) client works
 day one — and adds its own faster **`punktfunk/1`** protocol that breaks the ~1 Gbps FEC wall with a
 **GF(2¹⁶) Leopard-RS** transport. A single shared **Rust core** (`punktfunk-core`) holds the
-protocol, FEC, and crypto, linked into the host and every client over a stable C ABI.
+protocol, FEC, and crypto, linked into the host and every native client — directly as a Rust crate
+on Linux and Windows, and over a stable C ABI from the Apple and Android apps.
 
 ## What makes it different
 
@@ -58,12 +59,16 @@ protocol, FEC, and crypto, linked into the host and every client over a stable C
 | **GameStream host** → stock Moonlight | ✅ Live end-to-end: pairing, RTSP, audio, per-client virtual output at native resolution, GPU zero-copy NVENC, gamepads |
 | **Native protocol** — `punktfunk/1` | ✅ Validated live: QUIC control + GF(2¹⁶) FEC/AES-GCM data plane, PIN pairing, mDNS discovery, mid-stream mode renegotiation |
 | **Windows host** (Windows 11 22H2+, x64) | 🟡 Implemented & shipping as a signed installer: its own all-Rust IddCx **virtual display** (secure-desktop capable) with a **sealed IDD-push** capture path — finished frames pushed straight into its own driver, not screen-scraped (no DDA/WGC) · GPU encode (NVENC on NVIDIA, AMF/QSV on AMD/Intel, software H.264 without a GPU) · WASAPI audio · bundled virtual-gamepad drivers (no ViGEmBus) · HDR incl. Vulkan-game HDR. NVIDIA live-validated; AMD/Intel CI-green |
-| **macOS / iOS / tvOS client** (`clients/apple`) | ✅ Streaming live: VideoToolbox decode, controllers incl. DualSense, discovery, pairing, speed test |
-| **Linux client** (`clients/linux`, GTK4) | ✅ Streaming live: FFmpeg + VAAPI zero-copy decode, PipeWire audio, SDL3 controllers; ships as Flatpak/apt/rpm/Arch |
+| **macOS / iOS / tvOS client** (`clients/apple`) | ✅ Streaming live: VideoToolbox decode (HEVC, and AV1 on hardware that decodes it), controllers incl. DualSense, discovery, pairing, speed test |
+| **Linux client** (`clients/linux` + `clients/session`) | ✅ Streaming live: relm4/GTK4 launcher shell that spawns a Vulkan session binary — Vulkan Video / VAAPI / software decode, PipeWire audio, SDL3 controllers, Skia console UI; ships as Flatpak/apt/rpm/Arch |
 | **Android client** (`clients/android`, phone + TV) | ✅ Streaming live: AMediaCodec decode + HDR10, AAudio audio, controllers, discovery, pairing |
-| **Windows client** (`clients/windows`, WinUI 3) | ✅ Streaming live: D3D11VA hardware decode on all GPU vendors (NVIDIA + Intel validated on glass) with software fallback, WASAPI audio, SDL3 controllers, discovery, pairing; ships as signed MSIX (x64 + ARM64). HDR10 implemented, on-glass validation pending |
+| **Windows client** (`clients/windows`, WinUI 3) | ✅ Streaming live: WinUI 3 shell + Vulkan session presenter, hardware decode on all GPU vendors via Vulkan Video → D3D11VA → software (NVIDIA + Intel validated on glass), WASAPI audio, SDL3 controllers, discovery, pairing; ships as signed MSIX (x64 + ARM64). HDR10 implemented, on-glass validation pending |
 | **Web console + management API** (`web/`) | ✅ TanStack console over the OpenAPI mgmt API: host status, paired devices, on-demand PIN pairing, GPU selection, performance capture graphs, live host logs |
 
+Every native client also ships a tiered **stats overlay** (Compact / Normal / Detailed) with a
+shared vocabulary across platforms, and the session client carries a full gamepad-driven **console
+shell** (`pf-console-ui`): host list, PIN pairing, settings, and an on-screen keyboard.
+
 The **GameStream host works with a stock Moonlight client** — validated live on NVIDIA hardware
 (RTX 5070 Ti, RTX 4090): PIN pairing that persists across restarts, an app catalog, RTSP/ENet/audio,
 and **video at the client's exact resolution and refresh** via a per-session virtual output (KWin,
@@ -117,7 +122,7 @@ Each client discovers hosts on the network automatically and does a one-time
 For development, or as an install fallback where no package is available:
 
 ```sh
-cargo build --workspace          # the Rust core, host, Linux client, and probe (Linux & macOS)
+cargo build --workspace          # core, host, tray, shared client crates, Linux shell + session client, probe (Linux & macOS)
 cargo test  --workspace          # unit + loopback + proptest + C ABI harness
 cargo clippy --workspace --all-targets -- -D warnings
 cargo fmt --all --check
@@ -137,18 +142,27 @@ and the [docs site](https://docs.punktfunk.unom.io).
 crates/
   punktfunk-core/   protocol · FEC · pacing · crypto · QUIC control plane — the C ABI (lib + cdylib + staticlib)
   punktfunk-host/   the host (Linux + Windows): virtual displays · capture · encode · input · GameStream · punktfunk/1 · mgmt
+  pf-client-core/   shared client plumbing (Linux + Windows): session pump · FFmpeg decode · audio · SDL3 gamepads · trust · discovery
+  pf-presenter/     Vulkan session presenter: SDL3 window · ash swapchain · frame present · input capture
+  pf-console-ui/    Skia console UI for the session client: gamepad shell · stats OSD · pairing · on-screen keyboard
+  pf-ffvk/          FFmpeg Vulkan hwcontext bindings (AVVkFrame) for Vulkan Video decode on the presenter's device
+  pf-driver-proto/  host ↔ pf-vdisplay driver contract: control IOCTLs + IDD-push frame transport (no_std)
+  punktfunk-tray/   host tray icon (Windows notification area / Linux StatusNotifierItem)
 clients/
   apple/    macOS / iOS / tvOS app (Swift · VideoToolbox · Metal · GameController)
-  linux/    Linux desktop app (Rust · GTK4/libadwaita · FFmpeg/VAAPI · PipeWire · SDL3)
+  linux/    Linux launcher shell (Rust · relm4 / GTK4 / libadwaita) — spawns the session client to stream
+  session/  punktfunk-session, the Vulkan streaming session (Rust · SDL3 · ash · Skia console UI) — also runs standalone (gamescope, Decky)
   windows/  Windows desktop app (Rust · WinUI 3 · D3D11 · WASAPI · SDL3)
   android/  Android phone + TV app (Kotlin · Rust JNI core · AMediaCodec · AAudio)
   probe/    headless reference / measurement client for punktfunk/1
   decky/    Steam Deck Decky plugin
 web/                         web console (TanStack) over the management API — status · devices · pairing · GPUs · performance · logs
+api/openapi.json             management-API OpenAPI spec (regenerated via `punktfunk-host openapi`, checked in)
 packaging/                   apt · rpm / COPR · Arch · Flatpak · Bazzite bootc image
 docs-site/                   public documentation site (Fumadocs) — https://docs.punktfunk.unom.io
 include/punktfunk_core.h     cbindgen-generated C header (checked in)
 tools/                       latency-probe · loss-harness (measurement)
+ci/                          CI container images (rust-ci · fedora-rpm)
 ```
 
 ## Design invariants

SECURITY.md

@@ -14,7 +14,7 @@ exposes other users before a fix exists.
 
 The more of this you can give us, the faster we can act:
 
-- The component and version (e.g. `punktfunk-host 0.6.0`, Windows or Linux, which client).
+- The component and version (e.g. `punktfunk-host 0.9.0`, Windows or Linux, which client).
 - The impact — what an attacker can do, and from what position (same LAN, a local service account,
   admin, a paired client, …).
 - Steps to reproduce, a proof-of-concept, or a crash/log if you have one.

api/openapi.json

@@ -10,7 +10,7 @@
       "name": "MIT OR Apache-2.0",
       "identifier": "MIT OR Apache-2.0"
     },
-    "version": "0.9.0"
+    "version": "0.9.1"
   },
   "paths": {
     "/api/v1/clients": {

clients/android/README.md

@@ -58,7 +58,7 @@ kit/              :kit — NativeBridge · native mDNS discovery · Gamepad · K
 `build-tools;37.0.0`, **`cmake;3.22.1`** (builds libopus); **JDK 21** (AGP 9.2 runs on JDK 17–21, not
 a newer default); Rust with `rustup target add aarch64-linux-android armv7-linux-androideabi x86_64-linux-android` and
 `cargo install cargo-ndk`. Toolchain is pinned (AGP 9.2 · Gradle 9.4.1 · Kotlin 2.3.21 · Compose BOM
-2026.05.01 · compileSdk 37 · minSdk 31).
+2026.05.01 · compileSdk 37 · minSdk 28).
 
 **Android Studio:** open `clients/android` — it uses its bundled JBR 21, and the `cargoNdk*` task
 builds the `.so` as part of the normal build.

clients/android/kit/build.gradle.kts

@@ -91,9 +91,12 @@ fun registerCargoNdk(taskName: String, release: Boolean) =
         val cmd = mutableListOf(
             "$cargoBin/cargo", "ndk",
             "-t", "arm64-v8a", "-t", "armeabi-v7a", "-t", "x86_64",
-            // Link against the minSdk-28 sysroot: libaaudio (API 26) is present, and building at the
-            // floor makes the linker reject any accidental >28 hard import (the one API-30 call we
-            // make, ANativeWindow_setFrameRate, is dlsym-resolved — see decode::try_set_frame_rate).
+            // Link against the minSdk-28 sysroot (libaaudio, API 26, is present). NOTE: this does
+            // NOT reject an accidental >28 hard import — a cdylib link permits undefined symbols,
+            // which then fail at System.loadLibrary on every device below the symbol's API level
+            // (the 0.9.0 Android-≤12 regression). The checkJniImports* task after this build is
+            // what actually enforces the floor; >28 entry points must be dlsym-resolved (see
+            // decode::try_set_frame_rate, decode::install_render_callback, adpf).
             "--platform", "28",
             "-o", file("src/main/jniLibs").absolutePath,
             "build", "-p", "punktfunk-client-android",
@@ -102,8 +105,28 @@ fun registerCargoNdk(taskName: String, release: Boolean) =
         commandLine(cmd)
     }
 
+// Post-link floor check: every undefined symbol in the built .so must exist in the API-28 stubs,
+// else System.loadLibrary fails on devices at the minSdk floor (see the script header for the
+// 0.9.0 incident this guards against). Runs right after its cargo-ndk task; the APK build depends
+// on this task (not the cargo one directly), so a violation fails the build, local and CI alike.
+fun registerCheckJniImports(taskName: String, cargoTask: TaskProvider<Exec>) =
+    tasks.register<Exec>(taskName) {
+        group = "rust"
+        description = "verify libpunktfunk_android.so imports stay within the API-28 floor"
+        dependsOn(cargoTask)
+        workingDir = repoRoot
+        commandLine(
+            "sh", File(repoRoot, "scripts/ci/check-android-jni-imports.sh").absolutePath,
+            "${androidSdkDir()}/ndk/$ndkVer",
+            file("src/main/jniLibs").absolutePath,
+            "28",
+        )
+    }
+
 val cargoNdkDebug = registerCargoNdk("cargoNdkDebug", release = false)
 val cargoNdkRelease = registerCargoNdk("cargoNdkRelease", release = true)
+val checkJniImportsDebug = registerCheckJniImports("checkJniImportsDebug", cargoNdkDebug)
+val checkJniImportsRelease = registerCheckJniImports("checkJniImportsRelease", cargoNdkRelease)
 
 afterEvaluate {
     // `-PskipRustBuild` skips the cargo-ndk native build — for JVM-only tasks (the Roborazzi
@@ -120,8 +143,9 @@ afterEvaluate {
     // debug build); only the cargo profile changes. `-PrustDebug` restores a debug-profile native
     // build for the rare session that actually steps through Rust.
     if (!project.hasProperty("skipRustBuild")) {
-        val debugRust = if (project.hasProperty("rustDebug")) cargoNdkDebug else cargoNdkRelease
+        val debugRust =
+            if (project.hasProperty("rustDebug")) checkJniImportsDebug else checkJniImportsRelease
         tasks.named("preDebugBuild").configure { dependsOn(debugRust) }
-        tasks.named("preReleaseBuild").configure { dependsOn(cargoNdkRelease) }
+        tasks.named("preReleaseBuild").configure { dependsOn(checkJniImportsRelease) }
     }
 }

clients/android/kit/src/main/kotlin/io/unom/punktfunk/kit/Gamepad.kt

@@ -175,6 +175,12 @@ object Gamepad {
      * Holds the previous axis/hat state so an unchanged frame emits nothing. One instance per
      * session; call [reset] on release-all (focus loss / disconnect / session stop) so nothing
      * sticks on the host (which has no client-side held-state knowledge).
+     *
+     * Single-source: only ONE qualifying controller feeds pad 0. Events must come from a device
+     * whose source classes include GAMEPAD (see [onMotion]) and the mapper pins itself to the
+     * first such device — a controller's joystick-classified sibling nodes (DualSense/DS4 motion
+     * sensors) and any second pad report every axis as 0, and folding them into the same state
+     * flapped a held trigger/stick between its value and 0 on every event interleave.
      */
     class AxisMapper(private val handle: Long) {
         // Sentinel so the first real value (incl. 0) always sends once after attach (Linux parity).
@@ -182,10 +188,29 @@ object Gamepad {
         private var hatX = 0 // -1 / 0 / +1
         private var hatY = 0
 
+        /** deviceId of the controller pad 0 is pinned to; −1 until the first qualifying event. */
+        private var deviceId = -1
+
         /** Returns true if this was a joystick ACTION_MOVE we consumed. */
         fun onMotion(event: MotionEvent): Boolean {
             if (!event.isFromSource(InputDevice.SOURCE_JOYSTICK)) return false
             if (event.actionMasked != MotionEvent.ACTION_MOVE) return false
+            // Only a true gamepad drives pad 0. A joystick ACTION_MOVE's own source is plain
+            // JOYSTICK for every sender, so qualify by the DEVICE's source classes: a real pad
+            // carries the GAMEPAD (button) class too, its sensor/touchpad sibling nodes and
+            // joystick-class remotes don't — and those report every pad axis as 0 (see the
+            // class doc for the held-trigger flap this caused).
+            val dev = event.device ?: return false
+            if (dev.sources and InputDevice.SOURCE_GAMEPAD != InputDevice.SOURCE_GAMEPAD) return false
+            // Single-pad model: pin to the first qualifying controller so a second pad (or its
+            // stick drift) can't fight pad 0; re-adopt only once the pinned device is gone.
+            if (deviceId != event.deviceId) {
+                if (deviceId != -1) {
+                    if (InputDevice.getDevice(deviceId) != null) return false
+                    reset() // the pinned pad is gone — lift its held state before adopting
+                }
+                deviceId = event.deviceId
+            }
 
             // Sticks: Android floats −1..1, +y = down → ±32767, negate Y for the wire's +y = up.
             sendAxis(AXIS_LS_X, stick(event.getAxisValue(MotionEvent.AXIS_X)))
@@ -193,9 +218,27 @@ object Gamepad {
             sendAxis(AXIS_RS_X, stick(event.getAxisValue(MotionEvent.AXIS_Z)))
             sendAxis(AXIS_RS_Y, stick(-event.getAxisValue(MotionEvent.AXIS_RZ)))
 
-            // Triggers: LTRIGGER/RTRIGGER if present, else BRAKE/GAS; 0..1 float → 0..255.
-            sendAxis(AXIS_LT, trigger(firstNonZero(event, MotionEvent.AXIS_LTRIGGER, MotionEvent.AXIS_BRAKE)))
-            sendAxis(AXIS_RT, trigger(firstNonZero(event, MotionEvent.AXIS_RTRIGGER, MotionEvent.AXIS_GAS)))
+            // Triggers: pads report LTRIGGER/RTRIGGER or BRAKE/GAS (some mirror both) — merge
+            // with max, the same fold as the Controllers screen probe, so a pad that reports
+            // only one pair and a pad that reports both behave identically; 0..1 → 0..255.
+            sendAxis(
+                AXIS_LT,
+                trigger(
+                    maxOf(
+                        event.getAxisValue(MotionEvent.AXIS_LTRIGGER),
+                        event.getAxisValue(MotionEvent.AXIS_BRAKE),
+                    ),
+                ),
+            )
+            sendAxis(
+                AXIS_RT,
+                trigger(
+                    maxOf(
+                        event.getAxisValue(MotionEvent.AXIS_RTRIGGER),
+                        event.getAxisValue(MotionEvent.AXIS_GAS),
+                    ),
+                ),
+            )
 
             // HAT → dpad button transitions (track previous, emit only the deltas).
             val hx = sign(event.getAxisValue(MotionEvent.AXIS_HAT_X))
@@ -237,10 +280,5 @@ object Gamepad {
         private fun trigger(v: Float): Int = (v.coerceIn(0f, 1f) * 255f).toInt()
 
         private fun sign(v: Float): Int = if (v < -0.5f) -1 else if (v > 0.5f) 1 else 0
-
-        private fun firstNonZero(e: MotionEvent, a: Int, b: Int): Float {
-            val va = e.getAxisValue(a)
-            return if (va != 0f) va else e.getAxisValue(b)
-        }
     }
 }

clients/android/native/Cargo.toml

@@ -43,15 +43,20 @@ tracing = { version = "0.1", default-features = false, features = ["std", "log"]
 # Pure-Rust FFI to libmediandk/libnativewindow/libaaudio — no C++/libc++_shared to bundle. Decode +
 # audio run entirely in Rust on native threads (the "no async on the hot path" invariant).
 # api-level-28 matches the app's minSdk floor (Android 9). AAudio (26), AMediaCodec (21) and
-# ANativeWindow_setBuffersDataSpace (28) are all ≤28; the one API-30 call we make
-# (ANativeWindow_setFrameRate) is dlsym-resolved at runtime (see decode::try_set_frame_rate), not
-# linked, so the .so still loads on API 28/29.
+# ANativeWindow_setBuffersDataSpace (28) are all ≤28; every call above the floor —
+# ANativeWindow_setFrameRate (30), …WithChangeStrategy (31), AMediaCodec_setOnFrameRenderedCallback
+# (33), the ADPF hints — is dlsym-resolved at runtime (decode::try_set_frame_rate,
+# decode::install_render_callback, adpf), never linked, so the .so still loads on API 28+.
 ndk = { version = "0.9", features = ["media", "audio", "nativewindow", "api-level-28"] }
-# Raw FFI for the one AMediaCodec entry point the `ndk` wrapper lacks:
-# `AMediaCodec_setOnFrameRenderedCallback` (API 26, under the minSdk-28 floor ⇒ hard-linked) — the
-# per-frame render-timestamp callback behind the HUD's `display` stage (see decode::DisplayTracker).
-# Reaching it needs the codec's raw pointer, which is why the workspace pins `ndk` to the vendored
-# copy whose only patch makes `MediaCodec::as_ptr` public (vendor/ndk, wired in the root Cargo.toml).
+# Raw FFI *types* for the one AMediaCodec entry point the `ndk` wrapper lacks:
+# `AMediaCodec_setOnFrameRenderedCallback` — the per-frame render-timestamp callback behind the
+# HUD's `display` stage (see decode::DisplayTracker). The symbol is API 33 ("Available since
+# Android T"), ABOVE the minSdk-28 floor, so it is dlsym-resolved at runtime
+# (decode::install_render_callback), NEVER hard-linked: 0.9.0 hard-linked it and
+# `System.loadLibrary` failed on every pre-Android-13 device. Only ndk-sys's pointer/typedef
+# types are referenced, which creates no import. Reaching it needs the codec's raw pointer, which
+# is why the workspace pins `ndk` to the vendored copy whose only patch makes `MediaCodec::as_ptr`
+# public (vendor/ndk, wired in the root Cargo.toml).
 ndk-sys = { version = "0.6", features = ["media"] }
 # setpriority/gettid to raise the decode thread toward URGENT_DISPLAY (see decode::boost_thread_priority).
 libc = "0.2"

clients/android/native/src/decode.rs

@@ -449,26 +449,49 @@ impl DisplayTracker {
     }
 }
 
-/// Register [`on_frame_rendered`] on the codec (`AMediaCodec_setOnFrameRenderedCallback`, API 26 —
-/// under the minSdk-28 floor, so hard-linked via `ndk-sys`; the `ndk` wrapper has no binding, which
-/// is what the vendored crate's public `as_ptr` patch is for). Returns the userdata pointer holding
-/// a leaked `Arc<DisplayTracker>` refcount; the caller MUST reclaim it with
-/// [`release_render_callback`] AFTER dropping the codec (`AMediaCodec_delete` is what guarantees no
-/// further callback can fire). `None` (nothing to reclaim) if the platform refused — the HUD then
-/// simply has no `display` stage, exactly the pre-callback behaviour.
+/// Register [`on_frame_rendered`] on the codec (`AMediaCodec_setOnFrameRenderedCallback`,
+/// **API 33** — "Available since Android T" per the NDK header; only the *Java* listener dates
+/// back further). That sits above the API-28 floor, so the entry point is dlsym-resolved at
+/// runtime like [`try_set_frame_rate`] — hard-linking it (as 0.9.0 shipped) made
+/// `System.loadLibrary` fail on every pre-Android-13 device, taking down all of `NativeBridge`.
+/// The `ndk` wrapper has no binding and the call needs the raw codec pointer, which is what the
+/// vendored crate's public `as_ptr` patch is for. Returns the userdata pointer holding a leaked
+/// `Arc<DisplayTracker>` refcount; the caller MUST reclaim it with [`release_render_callback`]
+/// AFTER dropping the codec (`AMediaCodec_delete` is what guarantees no further callback can
+/// fire). `None` (nothing to reclaim) if the symbol is absent (API < 33) or the platform refused —
+/// the HUD then simply has no `display` stage, exactly the pre-callback behaviour.
 fn install_render_callback(
     codec: &MediaCodec,
     tracker: &Arc<DisplayTracker>,
 ) -> Option<*const DisplayTracker> {
+    // media_status_t AMediaCodec_setOnFrameRenderedCallback(
+    //     AMediaCodec*, AMediaCodecOnFrameRendered, void*)                            (API 33)
+    type SetOnFrameRenderedFn = unsafe extern "C" fn(
+        *mut ndk_sys::AMediaCodec,
+        ndk_sys::AMediaCodecOnFrameRendered,
+        *mut c_void,
+    ) -> ndk_sys::media_status_t;
+    // SAFETY: `dlopen` of `libmediandk.so`, which the `ndk` media wrapper already links — always
+    // mapped, so this only bumps its refcount (never closed — process-lifetime handle). `dlsym`
+    // returns null when the symbol is absent (device below API 33), checked before transmuting the
+    // non-null pointer to its fn-pointer type.
+    let set_on_frame_rendered = unsafe {
+        let lib = libc::dlopen(c"libmediandk.so".as_ptr(), libc::RTLD_NOW);
+        if lib.is_null() {
+            return None;
+        }
+        let sym = libc::dlsym(lib, c"AMediaCodec_setOnFrameRenderedCallback".as_ptr());
+        if sym.is_null() {
+            log::info!("decode: no render callback on this API level (<33) — no display stage");
+            return None;
+        }
+        std::mem::transmute::<*mut c_void, SetOnFrameRenderedFn>(sym)
+    };
     let ud = Arc::into_raw(tracker.clone());
     // SAFETY: `codec.as_ptr()` is the live codec this thread owns; `ud` outlives the registration
     // (reclaimed only after the codec is deleted, per this function's contract).
     let status = unsafe {
-        ndk_sys::AMediaCodec_setOnFrameRenderedCallback(
-            codec.as_ptr(),
-            Some(on_frame_rendered),
-            ud as *mut c_void,
-        )
+        set_on_frame_rendered(codec.as_ptr(), Some(on_frame_rendered), ud as *mut c_void)
     };
     if status == ndk_sys::media_status_t::AMEDIA_OK {
         Some(ud)

clients/android/native/vendor/ndk/README.punktfunk.md

@@ -7,7 +7,7 @@ in the workspace root.
 **The only change** is in `src/media/media_codec.rs`: `MediaCodec::as_ptr` is made
 `pub` (upstream keeps it private) so the Android client can register
 `AMediaCodec_setOnFrameRenderedCallback` through `ndk-sys` — the render-timestamp
-callback behind the HUD's `display` stage (`design/stats-unification.md`), which the
+callback behind the HUD's `display` stage (punktfunk-planning: `stats-unification.md`), which the
 wrapper doesn't expose. Grep for `punktfunk vendored patch` to find it.
 
 Drop this vendor copy when upstream exposes the raw pointer or a frame-rendered

clients/android/native/vendor/ndk/src/input_queue.rs

@@ -130,7 +130,7 @@ impl InputQueue {
                 looper.ptr().as_ptr(),
                 id,
                 None,
-                std::ptr::null_mut(),
+                ptr::null_mut(),
             )
         }
     }

clients/android/native/vendor/ndk/src/media/media_format.rs

@@ -9,7 +9,11 @@ use std::{
     slice,
 };
 
-use crate::media_error::{MediaError, Result};
+use crate::media_error::MediaError;
+// `Result` is only referenced by the api-level-29 methods below; an ungated import warns
+// (unused_imports) on every default-feature build.
+#[cfg(feature = "api-level-29")]
+use crate::media_error::Result;
 
 /// A native [`AMediaFormat *`]
 ///

clients/apple/README.md

@@ -122,7 +122,8 @@ PUNKTFUNK_AUTOCONNECT=<box-ip> PUNKTFUNK_MODE=1280x720x60 swift run PunktfunkCli
 - **App Store screenshots** are automated — `tools/screenshots.sh all` renders the real UI at the
   required pixel sizes via a DEBUG-only shot mode; the `apple` CI workflow captures the iOS sizes on
   every main push. See the script header for details.
-- Deeper design notes live in [`design/apple-stage2-presenter.md`](../../design/apple-stage2-presenter.md).
+- Deeper design notes live in the internal planning repo (punktfunk-planning:
+  `apple-stage2-presenter.md`).
 
 ## Related
 

clients/apple/Sources/PunktfunkClient/ContentView.swift

@@ -31,8 +31,15 @@ struct ContentView: View {
     @AppStorage(DefaultsKey.codec) private var codec = "auto"
     @AppStorage(DefaultsKey.hdrEnabled) private var hdrEnabled = true
     @AppStorage(DefaultsKey.fullscreenWhileStreaming) private var fullscreenWhileStreaming = true
-    @AppStorage(DefaultsKey.hudEnabled) private var hudEnabled = true
+    // The raw string is what @AppStorage observes (so cycles from any surface re-render this
+    // view); the absent-key default runs the legacy-hudEnabled migration once per init.
+    @AppStorage(DefaultsKey.statsVerbosity) private var statsVerbosityRaw
+        = StatsVerbosity.current.rawValue
     @AppStorage(DefaultsKey.hudPlacement) private var hudPlacement = HUDPlacement.topTrailing.rawValue
+    /// The persisted overlay tier (unknown raw falls back to .normal, like the migration).
+    private var statsVerbosity: StatsVerbosity {
+        StatsVerbosity(rawValue: statsVerbosityRaw) ?? .normal
+    }
     /// The `codec` setting as a `PUNKTFUNK_CODEC_*` soft-preference byte (`0` = auto).
     private var preferredCodecByte: UInt8 {
         switch codec {
@@ -393,9 +400,23 @@ struct ContentView: View {
                     displayMeter: model.displayStage
                 )
                 .overlay(alignment: placement.alignment) {
-                    if captureEnabled && hudEnabled {
-                        StreamHUDView(model: model, connection: conn, placement: placement)
+                    // The stats overlay MORPHS between tiers and SCALES UP on enter. With no `.id`, a
+                    // verbosity change keeps the same StreamHUDView identity, so its one shared glass
+                    // card animates its frame/shape to the new tier (a morph) instead of cross-fading a
+                    // fresh card in. The `.transition` therefore fires only on the off↔on boundary — a
+                    // scale-up (0.8→1) from the HUD's own corner. The ZStack is the stable host the
+                    // `.animation` watches as the child enters/leaves and morphs.
+                    ZStack {
+                        if captureEnabled && statsVerbosity != .off {
+                            StreamHUDView(
+                                model: model, connection: conn, placement: placement,
+                                verbosity: statsVerbosity)
+                                .transition(
+                                    .scale(scale: 0.8, anchor: placement.unitPoint)
+                                        .combined(with: .opacity))
+                        }
                     }
+                    .animation(.smooth(duration: 0.28), value: statsVerbosity)
                 }
                 #if os(macOS)
                 // The start-of-stream shortcut banner (Windows-client parity): the full
@@ -422,17 +443,18 @@ struct ContentView: View {
                 }
                 #endif
                 #if os(iOS)
-                // Touch users have no menu / ⌘D, so when the HUD (and its Disconnect button)
-                // is hidden, keep a minimal always-reachable exit in a corner. It rides a
-                // material disc (like the HUD) so the glyph stays legible over a bright frame
-                // — this is the sole touch disconnect path when stats are off.
+                // Touch users have no menu / ⌘D, so when the HUD's Disconnect button isn't on
+                // screen — the overlay off, or the compact pill (which carries no button) —
+                // keep a minimal always-reachable exit in a corner. It rides a material disc
+                // (like the HUD) so the glyph stays legible over a bright frame — this is the
+                // sole touch disconnect path in those tiers.
                 .overlay(alignment: .topLeading) {
-                    if captureEnabled && !hudEnabled {
+                    if captureEnabled && (statsVerbosity == .off || statsVerbosity == .compact) {
                         Button { model.disconnect() } label: {
                             Image(systemName: "xmark")
                                 .font(.headline.weight(.semibold))
                                 .frame(width: 36, height: 36)
-                                // Sole touch exit when the HUD is off — a floating glass disc
+                                // Sole touch exit in the off/compact tiers — a floating glass disc
                                 // over the frame (26+, material fallback). interactive: the disc
                                 // IS the tap target, so the glass reacts to press.
                                 .glassBackground(Circle(), interactive: true)

clients/apple/Sources/PunktfunkClient/Session/HUDPlacement.swift

@@ -23,6 +23,17 @@ enum HUDPlacement: String, CaseIterable, Identifiable {
     /// The HUD's own stack hugs the screen edge it sits against, so its text aligns outward.
     var isTrailing: Bool { self == .topTrailing || self == .bottomTrailing }
 
+    /// The corner as a `UnitPoint`, so a scale transition grows/shrinks the HUD out of its own
+    /// corner rather than its centre.
+    var unitPoint: UnitPoint {
+        switch self {
+        case .topLeading: return .topLeading
+        case .topTrailing: return .topTrailing
+        case .bottomLeading: return .bottomLeading
+        case .bottomTrailing: return .bottomTrailing
+        }
+    }
+
     /// User-facing corner label.
     var label: String {
         switch self {

clients/apple/Sources/PunktfunkClient/Session/StreamCommands.swift

@@ -5,8 +5,9 @@
 // (the Linux client has its GTK Shortcuts window, Windows its start-of-stream banner). While
 // input is CAPTURED these key equivalents never reach the menu (the stream view swallows
 // keys); InputCapture's monitor detects the same combos there and performs the same actions —
-// the menu covers the released state and discoverability. The stats toggle just flips the
-// shared `hudEnabled` setting; ContentView reads the same @AppStorage and reacts.
+// the menu covers the released state and discoverability. The stats item cycles the shared
+// `statsVerbosity` tier (off → compact → normal → detailed → off); ContentView reads the same
+// @AppStorage and reacts.
 //
 // tvOS has no menu bar / hardware-keyboard command surface (disconnect there is the Siri
 // Remote's Menu button, handled by ContentView's `.onExitCommand`), so this whole file is
@@ -36,12 +37,16 @@ extension FocusedValues {
 
 struct StreamCommands: Commands {
     @FocusedValue(\.sessionFocus) private var session
-    @AppStorage(DefaultsKey.hudEnabled) private var hudEnabled = true
+    // The raw string so @AppStorage observes the shared key; the absent-key default runs the
+    // legacy-hudEnabled migration (same pattern as ContentView/SettingsView).
+    @AppStorage(DefaultsKey.statsVerbosity) private var statsVerbosityRaw
+        = StatsVerbosity.current.rawValue
 
     var body: some Commands {
         CommandMenu("Stream") {
-            Button(hudEnabled ? "Hide Statistics" : "Show Statistics") {
-                hudEnabled.toggle()
+            Button("Cycle Statistics") {
+                let current = StatsVerbosity(rawValue: statsVerbosityRaw) ?? .normal
+                statsVerbosityRaw = current.next().rawValue
             }
             .keyboardShortcut("s", modifiers: [.control, .option, .shift])
             // Reaches the key window's stream view via NotificationCenter — capture is view

clients/apple/Sources/PunktfunkClient/Session/StreamHUDView.swift

@@ -1,17 +1,81 @@
-// The streaming overlay HUD: mode + fps/throughput, the unified latency lines
-// (design/stats-unification.md — end-to-end headline + the stage equation under stage-2, the
-// capture→received headline under the stage-1 fallback), the loss counter, the platform input
-// hint, and disconnect.
+// The streaming overlay HUD, tiered by StatsVerbosity (the Android client's 3-tier semantics):
+//  * compact — one glass-pill line: fps · end-to-end p50 · throughput (+ loss when lossy);
+//  * normal — mode + fps/throughput, the unified latency HEADLINE (design/stats-unification.md
+//    — end-to-end under stage-2, capture→received under the stage-1 fallback), the loss
+//    counter, a capture hint (shown until input is captured), and disconnect;
+//  * detailed — everything normal has plus the stage equation line(s) under the headline.
+// `.off` never reaches this view (ContentView gates the overlay on the tier).
 
 import PunktfunkKit
 import SwiftUI
+#if canImport(UIKit)
+import UIKit
+#endif
 
 struct StreamHUDView: View {
     @ObservedObject var model: SessionModel
     let connection: PunktfunkConnection
     var placement: HUDPlacement = .topTrailing
+    let verbosity: StatsVerbosity
 
     var body: some View {
+        // .off is gated upstream (ContentView only mounts the HUD when the tier is on) —
+        // render nothing if it ever slips through.
+        if verbosity != .off {
+            // ONE shared glass card wraps the tier-dependent content, so a verbosity change MORPHS
+            // this card — its frame (and, on iOS, its clamped corner) animate to the new size — rather
+            // than cross-fading a whole new card in. Only the inner content switches per tier.
+            tierContent
+                .padding(10)
+                .glassBackground(cardShape)
+                .padding(edgeInset)
+        }
+    }
+
+    /// The tier-dependent content, unwrapped (the shared card in `body` supplies the padding +
+    /// glass background). Compact is a one-line pill; normal/detailed the full stack.
+    @ViewBuilder private var tierContent: some View {
+        if verbosity == .compact {
+            compactContent
+        } else {
+            fullContent
+        }
+    }
+
+    // MARK: - Compact tier
+
+    /// One line: `{fps} fps · {e2e p50} ms · {mbps} Mb/s`. The ms segment is the best available
+    /// latency headline (stage-2 end-to-end, else the stage-1 capture→received) and is omitted until
+    /// either is valid. Loss appends in the same quiet styling the full HUD's lost line uses.
+    private var compactContent: some View {
+        HStack(spacing: 6) {
+            Circle()
+                .fill(Color.accentColor)
+                .frame(width: 7, height: 7)
+            Text(compactLine)
+                .font(.system(.caption, design: .monospaced))
+            if model.lostFrames > 0 {
+                Text("· lost \(model.lostFrames)")
+                    .font(.system(.caption2, design: .monospaced))
+                    .foregroundStyle(.secondary)
+            }
+        }
+    }
+
+    private var compactLine: String {
+        var parts = ["\(model.fps) fps"]
+        if model.endToEndValid {
+            parts.append(String(format: "%.1f ms", model.endToEndP50Ms))
+        } else if model.hostNetworkValid {
+            parts.append(String(format: "%.1f ms", model.hostNetworkP50Ms))
+        }
+        parts.append(String(format: "%.1f Mb/s", model.mbps))
+        return parts.joined(separator: " · ")
+    }
+
+    // MARK: - Normal / detailed tiers
+
+    private var fullContent: some View {
         VStack(alignment: placement.isTrailing ? .trailing : .leading, spacing: 4) {
             HStack(spacing: 6) {
                 Circle()
@@ -26,11 +90,11 @@ struct StreamHUDView: View {
                 Text("end-to-end \(model.endToEndP50Ms, specifier: "%.1f") ms p50 · \(model.endToEndP95Ms, specifier: "%.1f") p95 · capture→on-glass\(model.endToEndSkewCorrected ? "" : " (same-host clock)")")
                     .font(.system(.caption2, design: .monospaced))
                     .foregroundStyle(.secondary)
-                // The equation: the stages tiling the headline interval (per-window p50s —
-                // they only approximately sum to the directly-measured total). With a host
-                // that reports per-AU timings (0xCF) the first term splits into host + network
-                // (phase 2); an old host keeps the combined term.
-                if model.hostNetworkValid && model.decodeValid && model.displayValid {
+                // The equation (detailed tier only): the stages tiling the headline interval
+                // (per-window p50s — they only approximately sum to the directly-measured
+                // total). With a host that reports per-AU timings (0xCF) the first term splits
+                // into host + network (phase 2); an old host keeps the combined term.
+                if verbosity == .detailed && model.hostNetworkValid && model.decodeValid && model.displayValid {
                     if model.splitValid {
                         Text("= host \(model.hostP50Ms, specifier: "%.1f") + network \(model.networkP50Ms, specifier: "%.1f") + decode \(model.decodeP50Ms, specifier: "%.1f") + display \(model.displayP50Ms, specifier: "%.1f")")
                             .font(.system(.caption2, design: .monospaced))
@@ -45,11 +109,12 @@ struct StreamHUDView: View {
                 // Stage-1 fallback presenter: the layer decodes + presents internally with no
                 // per-frame stamp, so the honest headline ends at receipt. The host/network
                 // split still applies there (receipt is presenter-independent) — it becomes the
-                // only equation line; without it, host+network IS the whole measured interval.
+                // only equation line (detailed tier); without it, host+network IS the whole
+                // measured interval.
                 Text("capture→received \(model.hostNetworkP50Ms, specifier: "%.1f") ms p50 · \(model.hostNetworkP95Ms, specifier: "%.1f") p95\(model.hostNetworkSkewCorrected ? "" : " (same-host clock)")")
                     .font(.system(.caption2, design: .monospaced))
                     .foregroundStyle(.secondary)
-                if model.splitValid {
+                if verbosity == .detailed && model.splitValid {
                     Text("= host \(model.hostP50Ms, specifier: "%.1f") + network \(model.networkP50Ms, specifier: "%.1f")")
                         .font(.system(.caption2, design: .monospaced))
                         .foregroundStyle(.secondary)
@@ -63,22 +128,22 @@ struct StreamHUDView: View {
                     .font(.system(.caption2, design: .monospaced))
                     .foregroundStyle(.secondary)
             }
-            // While captured the cursor is hidden+frozen, so the button is keyboard-only
-            // (⌃⌥⇧Q — the cross-client Ctrl+Alt+Shift+Q — or ⌘⎋/Cmd+Tab release the cursor;
-            // released, it's clickable again).
+            // Capture hint, shown only until input is captured — how to grab it. The RELEASE
+            // shortcut is intentionally not surfaced in the overlay (it lives on the Stream menu
+            // and, on macOS, the start-of-stream banner), keeping the HUD uncluttered while playing.
             #if os(macOS)
-            Text(model.mouseCaptured
-                ? "⌃⌥⇧Q releases the mouse"
-                : "Click the stream to capture input")
-                .font(.geist(11, relativeTo: .caption2))
-                .foregroundStyle(.secondary)
+            if !model.mouseCaptured {
+                Text("Click the stream to capture input")
+                    .font(.geist(11, relativeTo: .caption2))
+                    .foregroundStyle(.secondary)
+            }
             #elseif os(iOS)
-            // Touch always plays directly; ⌘⎋ (hardware keyboard) toggles kb/mouse.
-            Text(model.mouseCaptured
-                ? "⌘⎋ releases keyboard & mouse"
-                : "⌘⎋ captures keyboard & mouse")
-                .font(.geist(11, relativeTo: .caption2))
-                .foregroundStyle(.secondary)
+            // Touch always plays directly; ⌘⎋ (hardware keyboard) captures kb/mouse.
+            if !model.mouseCaptured {
+                Text("⌘⎋ captures keyboard & mouse")
+                    .font(.geist(11, relativeTo: .caption2))
+                    .foregroundStyle(.secondary)
+            }
             #endif
             #if os(tvOS)
             // No focusable control during play: a focusable button steals the controller's
@@ -100,10 +165,58 @@ struct StreamHUDView: View {
             #endif
             #endif
         }
-        .padding(10)
-        // Floating HUD over live video — the canonical Liquid-Glass overlay surface (26+);
-        // falls back to .regularMaterial below 26 (see GlassStyle).
-        .glassBackground(RoundedRectangle(cornerRadius: 10))
-        .padding(10)
+    }
+
+    // MARK: - Card metrics
+
+    /// The OUTER gap between the card and the screen edge. (Inner content padding stays a fixed 10.)
+    /// On iOS the card hugs a physically rounded display corner, so it sits a little further in and
+    /// pairs with a concentric corner radius (below); on macOS/tvOS windows the classic 10 reads fine.
+    private var edgeInset: CGFloat {
+        #if os(iOS)
+        return 14
+        #else
+        return 10
+        #endif
+    }
+
+    /// The card's corner radius. On iOS it's concentric with the physical display corner —
+    /// `displayCornerRadius − edgeInset`, so the gap to the screen edge stays uniform right around the
+    /// corner instead of a small-radius card cutting into the very rounded glass. Clamped so a
+    /// flat-cornered device (or a hidden radius) still gets a sensibly rounded card.
+    private var cardCornerRadius: CGFloat {
+        #if os(iOS)
+        return max(12, DeviceMetrics.displayCornerRadius - edgeInset)
+        #else
+        return 10
+        #endif
+    }
+
+    /// The card background shape — a continuous (squircle) rounded rectangle, matching the curve
+    /// Apple's hardware display corners use so the concentric inset actually reads as parallel.
+    private var cardShape: RoundedRectangle {
+        RoundedRectangle(cornerRadius: cardCornerRadius, style: .continuous)
     }
 }
+
+#if os(iOS)
+/// Device display geometry the overlay needs but UIKit doesn't expose publicly.
+enum DeviceMetrics {
+    /// The physical display's corner radius. There's no public API for it, so read the private
+    /// `_displayCornerRadius` via KVC on the active window scene's screen, guarded by a fallback that
+    /// approximates a modern rounded device — a future OS that hides the key just yields a slightly
+    /// less-perfect inset, never a crash. The key is assembled from parts so it isn't a plain literal
+    /// in the binary; note the App Store private-API consideration regardless.
+    static var displayCornerRadius: CGFloat {
+        let key = ["_display", "Corner", "Radius"].joined()
+        guard
+            let screen = UIApplication.shared.connectedScenes
+                .compactMap({ $0 as? UIWindowScene })
+                .first?.screen,
+            let radius = screen.value(forKey: key) as? NSNumber,
+            radius.doubleValue > 0
+        else { return 44 }
+        return CGFloat(radius.doubleValue)
+    }
+}
+#endif

clients/apple/Sources/PunktfunkClient/Settings/GamepadSettingsView.swift

@@ -29,7 +29,10 @@ struct GamepadSettingsView: View {
     @AppStorage(DefaultsKey.enable444) private var enable444 = true
     @AppStorage(DefaultsKey.codec) private var codec = "auto"
     @AppStorage(DefaultsKey.micEnabled) private var micEnabled = true
-    @AppStorage(DefaultsKey.hudEnabled) private var hudEnabled = true
+    // The overlay tier's raw string (rows tag by rawValue); the absent-key default runs the
+    // legacy-hudEnabled migration (same pattern as ContentView/SettingsView).
+    @AppStorage(DefaultsKey.statsVerbosity) private var statsVerbosityRaw
+        = StatsVerbosity.current.rawValue
     @AppStorage(DefaultsKey.hudPlacement) private var hudPlacement = HUDPlacement.topTrailing.rawValue
     @AppStorage(DefaultsKey.libraryEnabled) private var libraryEnabled = false
     @AppStorage(DefaultsKey.gamepadUIEnabled) private var gamepadUIEnabled = true
@@ -271,10 +274,12 @@ struct GamepadSettingsView: View {
                 options: SettingsOptions.padTypes, current: gamepadType
             ) { gamepadType = $0 },
 
-            toggleRow(
+            choiceRow(
                 id: "hud", header: "Interface", icon: "chart.bar", label: "Statistics overlay",
-                detail: "Resolution, frame rate, throughput and latency while streaming.",
-                value: $hudEnabled),
+                detail: "How much to show while streaming — Compact is a one-line pill, "
+                    + "Detailed adds the latency stage breakdown.",
+                options: SettingsOptions.statsVerbosities, current: statsVerbosityRaw
+            ) { statsVerbosityRaw = $0 },
             choiceRow(
                 id: "hudPlacement", icon: "rectangle.inset.topright.filled", label: "Overlay position",
                 detail: "Which corner the statistics overlay sits in.",

clients/apple/Sources/PunktfunkClient/Settings/SettingsOptions.swift

@@ -37,6 +37,10 @@ enum SettingsOptions {
     static let hudPlacements: [(label: String, tag: String)] =
         HUDPlacement.allCases.map { ($0.label, $0.rawValue) }
 
+    /// Stats-overlay tiers (`DefaultsKey.statsVerbosity`) — the `tag` is the raw value.
+    static let statsVerbosities: [(label: String, tag: String)] =
+        StatsVerbosity.allCases.map { ($0.label, $0.rawValue) }
+
     /// Video-codec preference (`DefaultsKey.codec`) — a soft preference the host falls back from.
     /// AV1 appears only on devices with an AV1 hardware decoder (the same
     /// `AV1.hardwareDecodeSupported` gate SessionModel advertises by) — elsewhere it would be a

clients/apple/Sources/PunktfunkClient/Settings/SettingsView+Sections.swift

@@ -337,28 +337,32 @@ extension SettingsView {
         #endif
     }
 
-    // Stage-2 (Metal/VTDecompressionSession) is the default and only user-visible presenter — it
-    // recovers from a wedged decoder, where stage-1's AVSampleBufferDisplayLayer freezes hard on a
-    // lost HEVC reference. Stage-1 is kept reachable as a DEBUG-only override for diagnostics, like
-    // the controller test. Empty in release builds (no presenter UI; stage-2 always).
+    // Stage-2 (Metal/VTDecompressionSession, present on frame arrival) is the proven default;
+    // stage-3 is the same pipeline with glass-gated present pacing — a user-visible A/B while the
+    // pacing work settles (see Stage2Pipeline's PresentPacing for the queue-saturation rationale).
+    // Stage-1 (compressed video straight to the system layer) stays a DEBUG-only diagnostic — it
+    // freezes hard on a lost HEVC reference.
     @ViewBuilder var presenterSection: some View {
-        #if DEBUG
         Section {
             Picker("Presenter", selection: $presenter) {
                 Text("Stage 2 (default)").tag("stage2")
+                Text("Stage 3 (experimental)").tag("stage3")
+                #if DEBUG
                 Text("Stage 1 (debug)").tag("stage1")
+                #endif
             }
         } header: {
-            Text("Video presenter · debug")
+            Text("Video presenter")
         } footer: {
-            Text("Stage 2 (default): explicit decode + Metal present — full HUD latency "
-                + "breakdown and self-recovery from decode stalls. Stage 1: compressed video "
-                + "straight to the system layer; freezes on a lost HEVC reference, so it's a "
-                + "debug fallback only. Applies from the next session.")
+            Text("Stage 2: each frame is shown the moment it's decoded — proven, but on displays "
+                + "running near the stream's frame rate, queued frames can add two to three "
+                + "refreshes of display latency that never drains. Stage 3: presents are paced "
+                + "to the display — at most one undisplayed frame in flight, always the freshest, "
+                + "dropping late frames instead of queueing them. Watch the statistics overlay's "
+                + "display time to compare. Applies from the next session.")
                 .font(.geist(12, relativeTo: .caption))
                 .foregroundStyle(.secondary)
         }
-        #endif
     }
 
     @ViewBuilder var hdrSection: some View {
@@ -385,13 +389,17 @@ extension SettingsView {
 
     @ViewBuilder var statisticsSection: some View {
         Section {
-            Toggle("Show statistics overlay", isOn: $hudEnabled)
+            Picker("Statistics overlay", selection: $statsVerbosityRaw) {
+                ForEach(StatsVerbosity.allCases, id: \.rawValue) { tier in
+                    Text(tier.label).tag(tier.rawValue)
+                }
+            }
             Picker("Position", selection: $hudPlacement) {
                 ForEach(HUDPlacement.allCases) { placement in
                     Text(placement.label).tag(placement.rawValue)
                 }
             }
-            .disabled(!hudEnabled)
+            .disabled(statsVerbosityRaw == StatsVerbosity.off.rawValue)
         } header: {
             Text("Statistics")
         } footer: {

clients/apple/Sources/PunktfunkClient/Settings/SettingsView+Support.swift

@@ -54,10 +54,14 @@ extension SettingsView {
     // MARK: - Statistics
 
     static var statisticsFooter: String {
-        let base = "Shows resolution, frame rate, throughput and latency in the chosen "
-            + "corner while streaming."
-        #if os(macOS) || os(iOS)
-        return base + " Toggle it any time with ⌃⌥⇧S."
+        let base = "Shows streaming statistics in the chosen corner — Compact is a one-line "
+            + "pill, Normal adds resolution and latency, Detailed adds the latency stage "
+            + "breakdown."
+        #if os(macOS)
+        return base + " ⌃⌥⇧S cycles Off → Compact → Normal → Detailed any time."
+        #elseif os(iOS)
+        return base + " ⌃⌥⇧S or a three-finger tap cycles Off → Compact → Normal → Detailed "
+            + "any time."
         #else
         return base
         #endif

clients/apple/Sources/PunktfunkClient/Settings/SettingsView.swift

@@ -38,7 +38,9 @@ struct SettingsView: View {
     @AppStorage(DefaultsKey.micEnabled) var micEnabled = true
     @AppStorage(DefaultsKey.audioChannels) var audioChannels = 2
     @AppStorage(DefaultsKey.codec) var codec = "auto"
-    @AppStorage(DefaultsKey.hudEnabled) var hudEnabled = true
+    // The overlay tier's raw string (the pickers tag by rawValue); the absent-key default runs
+    // the legacy-hudEnabled migration (same pattern as ContentView/StreamCommands).
+    @AppStorage(DefaultsKey.statsVerbosity) var statsVerbosityRaw = StatsVerbosity.current.rawValue
     @AppStorage(DefaultsKey.hudPlacement) var hudPlacement = HUDPlacement.topTrailing.rawValue
     @ObservedObject var gamepads = GamepadManager.shared
     #if !os(tvOS)
@@ -282,6 +284,19 @@ struct SettingsView: View {
         ("4K @ 60", "3840x2160x60"),
     ]
 
+    /// Stage-2 vs stage-3 present pacing (see SettingsView+Sections' presenterSection for the
+    /// rationale); the freeze-prone stage-1 diagnostic only ships in DEBUG builds.
+    private static var presenterOptions: [(label: String, tag: String)] {
+        var options: [(label: String, tag: String)] = [
+            ("Stage 2 (default)", "stage2"),
+            ("Stage 3 (experimental)", "stage3"),
+        ]
+        #if DEBUG
+        options.append(("Stage 1 (debug)", "stage1"))
+        #endif
+        return options
+    }
+
     private var modeTag: Binding<String> {
         Binding(
             get: { "\(width)x\(height)x\(hz)" },
@@ -294,10 +309,6 @@ struct SettingsView: View {
             })
     }
 
-    private var hudEnabledTag: Binding<String> {
-        Binding(get: { hudEnabled ? "on" : "off" }, set: { hudEnabled = $0 == "on" })
-    }
-
     private var hdrEnabledTag: Binding<String> {
         Binding(get: { hdrEnabled ? "on" : "off" }, set: { hdrEnabled = $0 == "on" })
     }
@@ -334,12 +345,10 @@ struct SettingsView: View {
                 TVSelectionRow(
                     title: "Compositor", options: SettingsOptions.compositors,
                     selection: $compositor)
-                #if DEBUG
                 TVSelectionRow(
-                    title: "Presenter (debug)",
-                    options: [("Stage 2 (default)", "stage2"), ("Stage 1 (debug)", "stage1")],
+                    title: "Presenter",
+                    options: Self.presenterOptions,
                     selection: $presenter)
-                #endif
                 TVSelectionRow(
                     title: "10-bit HDR",
                     options: [("On", "on"), ("Off", "off")], selection: hdrEnabledTag)
@@ -352,7 +361,7 @@ struct SettingsView: View {
                     .padding(.top, 8)
                 TVSelectionRow(
                     title: "Statistics overlay",
-                    options: [("On", "on"), ("Off", "off")], selection: hudEnabledTag)
+                    options: SettingsOptions.statsVerbosities, selection: $statsVerbosityRaw)
                 TVSelectionRow(
                     title: "Statistics position", options: SettingsOptions.hudPlacements,
                     selection: $hudPlacement)

clients/apple/Sources/PunktfunkKit/Input/InputCapture.swift

@@ -113,11 +113,11 @@ public final class InputCapture {
     /// keyDown monitor only WHILE FORWARDING — that's the state in which the app's menu (which
     /// carries the same key equivalents for discoverability) can't see them, so the monitor is the
     /// captured-state delivery path; released, the events pass through and the menu handles them.
-    /// ⌃⌥⇧Q releases the captured mouse/keyboard; ⌃⌥⇧D disconnects; ⌃⌥⇧S toggles the stats
-    /// overlay. Main queue.
+    /// ⌃⌥⇧Q releases the captured mouse/keyboard; ⌃⌥⇧D disconnects; ⌃⌥⇧S cycles the stats
+    /// overlay tier (off → compact → normal → detailed). Main queue.
     public var onReleaseCapture: (() -> Void)?
     public var onDisconnect: (() -> Void)?
-    public var onToggleStats: (() -> Void)?
+    public var onCycleStats: (() -> Void)?
 
     /// Fired when a newer InputCapture takes the process-global GC handler slots (the
     /// singletons hold ONE handler each): the preempted owner must drop its capture
@@ -246,7 +246,7 @@ public final class InputCapture {
                     return nil
                 case 1 /* S */:
                     self.suppressedVK = 0x53
-                    self.onToggleStats?()
+                    self.onCycleStats?()
                     return nil
                 default:
                     break

clients/apple/Sources/PunktfunkKit/Input/TouchMouse.swift

@@ -2,7 +2,8 @@
 // touch gesture model (clients/android .../TouchInput.kt) so the two touch clients feel
 // identical. Two mouse modes share one gesture vocabulary — tap = left click · two-finger
 // tap = right click · two-finger drag = scroll · tap-then-press-and-drag = held left drag
-// (text selection / window moves) · three-finger tap = stats-HUD toggle:
+// (text selection / window moves) · three-finger tap = cycles the stats overlay tiers
+// (off → compact → normal → detailed, matching Android):
 //
 //  * trackpad (default): the cursor STAYS PUT on touch-down and moves by the finger's
 //    relative delta with mild acceleration — swipe to nudge, lift and re-swipe to walk it
@@ -164,7 +165,7 @@ final class TouchMouse {
         } else if !moved {
             switch maxFingers {
             case 3...:
-                Self.toggleHUD() // in-stream stats-overlay toggle, same as Android
+                Self.cycleStats() // in-stream stats-tier cycle, same as Android
             case 2: // two-finger tap → right click
                 send?(.mouseButton(Button.right, down: true))
                 send?(.mouseButton(Button.right, down: false))
@@ -274,12 +275,11 @@ final class TouchMouse {
         }
     }
 
-    /// Three-finger tap toggles the stats overlay — through the shared `hudEnabled` default,
-    /// which the app's HUD views observe via @AppStorage (so this needs no wiring to them).
-    private static func toggleHUD() {
-        let defaults = UserDefaults.standard
-        let on = defaults.object(forKey: DefaultsKey.hudEnabled) as? Bool ?? true
-        defaults.set(!on, forKey: DefaultsKey.hudEnabled)
+    /// Three-finger tap cycles the stats overlay tiers (off → compact → normal → detailed) —
+    /// through the shared `statsVerbosity` default, which the app's HUD views observe via
+    /// @AppStorage (so this needs no wiring to them). Same cycle as Android's triple-tap.
+    private static func cycleStats() {
+        StatsVerbosity.store(StatsVerbosity.current.next())
     }
 }
 #endif

clients/apple/Sources/PunktfunkKit/Support/DefaultsKeys.swift

@@ -30,6 +30,11 @@ public enum DefaultsKey {
     /// discrete channel, and the default N→stereo downmix grabs channels 0/1 (silence when the mic
     /// is higher up), so we fold to mono ourselves. Only meaningful for multi-channel devices.
     public static let micChannel = "punktfunk.micChannel"
+    /// Which presenter runs a session: "stage2" (default — explicit decode + Metal present on
+    /// frame arrival), "stage3" (same pipeline, glass-gated present pacing — the experimental
+    /// low-display-latency A/B; see Stage2Pipeline's PresentPacing), or "stage1" (DEBUG-only
+    /// system-layer fallback). Resolved once per session by SessionPresenter;
+    /// PUNKTFUNK_PRESENTER=stage1|stage2|stage3 overrides it for A/B.
     public static let presenter = "punktfunk.presenter"
     /// macOS: V-Sync the stream's presents — each decoded frame flips on the next display vsync
     /// (evenly paced, no tearing under direct scanout) instead of as soon as the GPU finishes
@@ -67,9 +72,15 @@ public enum DefaultsKey {
     public static let libraryEnabled = "punktfunk.libraryEnabled"
     /// macOS: take the window fullscreen while streaming and restore it on the host list. On by default.
     public static let fullscreenWhileStreaming = "punktfunk.fullscreenWhileStreaming"
-    /// Show the streaming statistics overlay (mode/fps/throughput/latency). On by default; toggle
-    /// while streaming with ⌃⌥⇧S (the cross-client Ctrl+Alt+Shift+S; macOS / hardware keyboard).
+    /// LEGACY (pre-tiered overlay): the old boolean stats-overlay toggle. Kept ONLY as the
+    /// migration fallback `StatsVerbosity.current` reads when `statsVerbosity` was never
+    /// written (absent-or-true → .normal, explicit false → .off). Never written anymore.
     public static let hudEnabled = "punktfunk.hudEnabled"
+    /// The statistics overlay tier — a `StatsVerbosity` raw value ("off"/"compact"/"normal"/
+    /// "detailed"). Absent → migrated from the legacy `hudEnabled` bool (see above). Cycle it
+    /// while streaming with ⌃⌥⇧S (the cross-client Ctrl+Alt+Shift+S; macOS / hardware
+    /// keyboard) or a three-finger tap (touch), matching the Android client.
+    public static let statsVerbosity = "punktfunk.statsVerbosity"
     /// Which corner the statistics overlay sits in — a `HUDPlacement` raw value
     /// ("topLeading"/"topTrailing"/"bottomLeading"/"bottomTrailing"). Default top-trailing.
     public static let hudPlacement = "punktfunk.hudPlacement"

clients/apple/Sources/PunktfunkKit/Support/StatsVerbosity.swift

@@ -0,0 +1,58 @@
+// The stats overlay's verbosity tier — a port of the Android client's 3-tier overlay semantics
+// so the two clients feel identical: Off → Compact (one-line pill) → Normal (headline stats) →
+// Detailed (plus the latency stage equation). Persisted under `DefaultsKey.statsVerbosity`;
+// the in-stream cycle surfaces (⌃⌥⇧S, the three-finger tap) advance it with
+// `store(current.next())`, and every UI reader observes the same default via @AppStorage.
+//
+// Lives in PunktfunkKit (not the app) because the kit's input paths (TouchMouse's three-finger
+// tap, InputCapture's captured-state ⌃⌥⇧S) cycle it directly.
+
+import Foundation
+
+/// How much of the streaming statistics overlay to show. The raw values are stable on disk —
+/// rename the cases freely, never the strings.
+public enum StatsVerbosity: String, CaseIterable, Sendable {
+    case off, compact, normal, detailed
+
+    /// User-facing tier label (Settings pickers, the gamepad settings row).
+    public var label: String {
+        switch self {
+        case .off: return "Off"
+        case .compact: return "Compact"
+        case .normal: return "Normal"
+        case .detailed: return "Detailed"
+        }
+    }
+
+    /// The next tier in the cycle: off → compact → normal → detailed → off (wrapping) —
+    /// the ⌃⌥⇧S / three-finger-tap order, same as Android.
+    public func next() -> StatsVerbosity {
+        switch self {
+        case .off: return .compact
+        case .compact: return .normal
+        case .normal: return .detailed
+        case .detailed: return .off
+        }
+    }
+
+    /// The persisted tier. When `statsVerbosity` has never been written, migrates from the
+    /// legacy `hudEnabled` bool the pre-tiered clients stored: absent-or-true → `.normal`
+    /// (the old "on" look minus the equation lines), explicit false → `.off`.
+    public static var current: StatsVerbosity {
+        let defaults = UserDefaults.standard
+        if let raw = defaults.string(forKey: DefaultsKey.statsVerbosity),
+           let tier = StatsVerbosity(rawValue: raw) {
+            return tier
+        }
+        if let legacy = defaults.object(forKey: DefaultsKey.hudEnabled) as? Bool, !legacy {
+            return .off
+        }
+        return .normal
+    }
+
+    /// Persist a tier (the cycle surfaces write through here; the Settings pickers write the
+    /// same key via @AppStorage).
+    public static func store(_ tier: StatsVerbosity) {
+        UserDefaults.standard.set(tier.rawValue, forKey: DefaultsKey.statsVerbosity)
+    }
+}

clients/apple/Sources/PunktfunkKit/Video/SessionPresenter.swift

@@ -22,6 +22,30 @@ public final class DisplayLinkProxy: NSObject {
     @objc public func tick(_ link: CADisplayLink) { onTick(link) }
 }
 
+/// Which presenter a session runs. Stage-2/stage-3 are the same Metal pipeline with arrival vs
+/// glass-gated present pacing (`PresentPacing` — see Stage2Pipeline for the tradeoff, and why
+/// stage-3 exists: stage-2's present-on-arrival saturates the layer's FIFO image queue on panels
+/// running near the stream rate). Stage-1 (compressed video straight to the system layer) is a
+/// DEBUG-only diagnostic. Internal (not private) for unit tests.
+enum PresenterChoice: Equatable {
+    case stage1
+    case stage2
+    case stage3
+
+    /// Resolve from the `PUNKTFUNK_PRESENTER` env override (A/B without touching settings) first,
+    /// then the persisted `DefaultsKey.presenter` setting; anything unknown (or an empty env var)
+    /// falls back to stage-2. `allowStage1` is false in release builds, where a leftover DEBUG
+    /// "stage1" value silently maps to stage-2 rather than reviving the freeze-prone fallback.
+    static func resolve(setting: String?, env: String?, allowStage1: Bool) -> PresenterChoice {
+        let raw = env.flatMap { $0.isEmpty ? nil : $0 } ?? setting
+        switch raw {
+        case "stage1": return allowStage1 ? .stage1 : .stage2
+        case "stage3": return .stage3
+        default: return .stage2
+        }
+    }
+}
+
 final class SessionPresenter {
     private var pump: StreamPump?
     private var stage2: Stage2Pipeline?
@@ -50,18 +74,24 @@ final class SessionPresenter {
 
         // Presenter choice — stage-2 is the DEFAULT (explicit VTDecompressionSession decode + a
         // CAMetalLayer/display-link present): it can detect + recover a wedged decoder where
-        // stage-1's AVSampleBufferDisplayLayer freezes hard on a lost HEVC reference. Stage-1 is
-        // reachable only via the DEBUG presenter toggle; release always takes stage-2 (the stage-1
-        // pump below stays the automatic fallback if Metal is missing).
+        // stage-1's AVSampleBufferDisplayLayer freezes hard on a lost HEVC reference. Stage-3 is
+        // the same pipeline with glass-gated present pacing (the settings picker's live A/B — see
+        // PresentPacing). Stage-1 is reachable only via the DEBUG presenter value; release maps it
+        // back to stage-2 (the stage-1 pump below stays the automatic fallback if Metal is missing).
         #if DEBUG
-        let forceStage1 = UserDefaults.standard.string(forKey: DefaultsKey.presenter) == "stage1"
+        let allowStage1 = true
         #else
-        let forceStage1 = false
+        let allowStage1 = false
         #endif
-        if !forceStage1,
+        let choice = PresenterChoice.resolve(
+            setting: UserDefaults.standard.string(forKey: DefaultsKey.presenter),
+            env: ProcessInfo.processInfo.environment["PUNKTFUNK_PRESENTER"],
+            allowStage1: allowStage1)
+        if choice != .stage1,
            let pipeline = Stage2Pipeline(
                endToEndMeter: endToEndMeter, decodeMeter: decodeMeter,
-               displayMeter: displayMeter) {
+               displayMeter: displayMeter,
+               pacing: choice == .stage3 ? .glass : .arrival) {
             let metal = pipeline.layer
             // The opaque metal layer composites OVER the AVSampleBufferDisplayLayer base, which
             // sits idle (un-enqueued) in stage-2. contentsScale + frame are set in layout().

clients/apple/Sources/PunktfunkKit/Video/Stage2Pipeline.swift

@@ -18,6 +18,10 @@
 //     – V-Sync ON: present(at: next vsync) predicted from the link's last phase/period, at most one
 //       period ahead by construction, falling back to immediate when the link data is stale — a
 //       schedule can never sit far in the future holding drawables hostage.
+// • Present PACING is the stage-2 vs stage-3 presenter split (`PresentPacing`, chosen per session
+//   by SessionPresenter from the presenter setting / PUNKTFUNK_PRESENTER): stage-2 presents on
+//   frame arrival; stage-3 additionally gates presents to ONE undisplayed drawable so the layer's
+//   FIFO image queue can never saturate — see PresentPacing's doc for the full rationale.
 // • Rendering lives on its own thread so any `nextDrawable()` wait lands off-main (input, SwiftUI).
 //
 // The render thread also stamps the unified latency stages (end-to-end capture→on-glass + decode and
@@ -98,6 +102,79 @@ private final class VsyncClock: @unchecked Sendable {
     }
 }
 
+/// When a ready frame is pushed to the layer — the stage-2 vs stage-3 presenter split. Same decode
+/// half, same newest-wins ring; only the present cadence differs.
+///
+/// - `arrival` (stage-2, the default): present the moment a frame is decoded. Lowest latency while
+///   the layer's image queue is shallow — but that queue is FIFO and consumed at one drawable per
+///   refresh (iOS always vsync-latches; the macOS 26 compositor latch-paces our out-of-band
+///   presents the same way when composited), so at stream rate ≈ refresh rate its depth is STICKY:
+///   one early burst (session start, a Wi-Fi clump) fills it to `maximumDrawableCount` and — with
+///   arrivals and latches then running at the same rate — it never drains. Every later frame rides
+///   ~2–3 refreshes of queue (the measured 29–30 ms display stage on 120 Hz ProMotion panels), and
+///   the full-queue regime is where host↔panel clock drift turns into periodic repeats/drops (the
+///   "fixed-interval" jitter reports).
+/// - `glass` (stage-3, experimental): at most ONE presented-but-undisplayed drawable in flight
+///   (`PresentGate`). The render thread presents only when the previous flip reached glass (the
+///   drawable's presented handler reopens the gate and re-signals); frames decoded meanwhile
+///   coalesce in the newest-wins ring. Freshness is preserved by DROPPING stale frames before
+///   present instead of queueing them behind the display — the hidden queue latency becomes
+///   explicit, correct frame drops.
+public enum PresentPacing: Sendable {
+    case arrival
+    case glass
+}
+
+/// Stage-3's present gate: admits one in-flight (presented, not yet on glass) drawable. The render
+/// thread `tryAcquire`s before taking a frame; the drawable's presented handler `release`s and
+/// re-signals the render thread. `staleAfter` is insurance against a present whose handler never
+/// fires (the macOS "out-of-band presents aren't damage" hazard class — see MetalVideoPresenter's
+/// init post-mortem): rather than freezing the stream, a stuck gate force-opens after 100 ms, a
+/// visible ~10 fps degradation that PUNKTFUNK_PRESENT_DEBUG's `forced` counter exposes (it reads 0
+/// on healthy systems). Internal (not private) for unit tests. Sendable; lock-guarded — the
+/// releaser runs on a Metal callback thread.
+final class PresentGate: @unchecked Sendable {
+    /// How long one pending present may hold the gate before it's presumed lost.
+    static let staleAfter: CFTimeInterval = 0.1
+
+    private let lock = NSLock()
+    private var pending = false
+    private var armedAt: CFTimeInterval = 0
+    private var forced = 0
+
+    /// Arm the gate for one present. False = a present is already in flight (and not stale) —
+    /// leave the frame in the ring; the presented handler's release/re-signal (or the next
+    /// display-link tick) retries with the freshest frame then.
+    func tryAcquire(now: CFTimeInterval) -> Bool {
+        lock.lock()
+        defer { lock.unlock() }
+        if pending {
+            guard now - armedAt > Self.staleAfter else { return false }
+            forced += 1 // presumed-lost present — reopen rather than stall the stream
+        }
+        pending = true
+        armedAt = now
+        return true
+    }
+
+    /// The in-flight present reached glass (or was dropped, or its render failed before a present
+    /// was registered) — reopen. Idempotent: a late stale-path double-release is harmless.
+    func release() {
+        lock.lock()
+        pending = false
+        lock.unlock()
+    }
+
+    /// Take-and-reset the force-open count (PUNKTFUNK_PRESENT_DEBUG's `forced` stat).
+    func drainForced() -> Int {
+        lock.lock()
+        defer { lock.unlock() }
+        let n = forced
+        forced = 0
+        return n
+    }
+}
+
 /// PUNKTFUNK_PRESENT_DEBUG=1 aggregation: one printed line per second from the render thread with
 /// the decode rate, render outcomes, the slowest render call (≈ nextDrawable wait) and the deltas
 /// between system-reported on-glass times (vsync-aligned presents show clean refresh-period
@@ -105,22 +182,39 @@ private final class VsyncClock: @unchecked Sendable {
 private final class PresentDebugStats: @unchecked Sendable {
     private let lock = NSLock()
     private var last = CACurrentMediaTime()
-    private var ok = 0, failed = 0, empty = 0, dropped = 0
+    private var ok = 0, failed = 0, empty = 0, dropped = 0, gated = 0
     private var maxRenderMs = 0.0
     private var lastGlassNs: Int64 = 0
     private var glassDeltasMs: [Double] = []
+    /// Presented-but-not-yet-on-glass drawables right now / the window's peak — the direct
+    /// measurement of the layer image-queue depth the stage-3 gate exists to bound (stage-2 on a
+    /// 120 Hz panel saturates this at ~maximumDrawableCount; stage-3 should peg it at 1).
+    private var inFlight = 0
+    private var maxInFlight = 0
 
     func emptyWake() { lock.lock(); empty += 1; lock.unlock() }
 
+    /// A wake that found the stage-3 gate closed (a present still in flight) — the frame stays in
+    /// the ring for the handler's re-signal. Includes display-link ticks while gated; a high count
+    /// is normal, it just shows the gate working.
+    func gatedWake() { lock.lock(); gated += 1; lock.unlock() }
+
     func renderReturned(ok rendered: Bool, tookMs: Double) {
         lock.lock()
-        if rendered { ok += 1 } else { failed += 1 }
+        if rendered {
+            ok += 1
+            inFlight += 1
+            maxInFlight = max(maxInFlight, inFlight)
+        } else {
+            failed += 1
+        }
         maxRenderMs = max(maxRenderMs, tookMs)
         lock.unlock()
     }
 
     func presented(atNs: Int64?) {
         lock.lock()
+        inFlight = max(0, inFlight - 1) // clamp: the handler can beat renderReturned's increment
         if let atNs {
             if lastGlassNs > 0 { glassDeltasMs.append(Double(atNs - lastGlassNs) / 1e6) }
             lastGlassNs = atNs
@@ -130,7 +224,7 @@ private final class PresentDebugStats: @unchecked Sendable {
         lock.unlock()
     }
 
-    func flushIfDue(ring: ReadyRing) {
+    func flushIfDue(ring: ReadyRing, gate: PresentGate?) {
         lock.lock()
         let now = CACurrentMediaTime()
         guard now - last >= 1 else { lock.unlock(); return }
@@ -139,12 +233,15 @@ private final class PresentDebugStats: @unchecked Sendable {
         let deltas = glassDeltasMs.sorted()
         let p50 = deltas.isEmpty ? 0 : deltas[deltas.count / 2]
         let dMax = deltas.last ?? 0
+        let inflightMax = maxInFlight
         let line = String(
-            format: "pf-present decoded=%d ok=%d fail=%d empty=%d dropped=%d "
-                + "maxRenderMs=%.1f glassDeltaMs p50=%.2f max=%.2f n=%d",
-            decoded, ok, failed, empty, dropped, maxRenderMs, p50, dMax, deltas.count)
-        ok = 0; failed = 0; empty = 0; dropped = 0
+            format: "pf-present decoded=%d ok=%d fail=%d empty=%d gated=%d dropped=%d "
+                + "maxRenderMs=%.1f inflightMax=%d forced=%d glassDeltaMs p50=%.2f max=%.2f n=%d",
+            decoded, ok, failed, empty, gated, dropped, maxRenderMs, inflightMax,
+            gate?.drainForced() ?? 0, p50, dMax, deltas.count)
+        ok = 0; failed = 0; empty = 0; dropped = 0; gated = 0
         maxRenderMs = 0
+        maxInFlight = inFlight // the window peak restarts from the live depth
         glassDeltasMs.removeAll(keepingCapacity: true)
         lock.unlock()
         print(line)
@@ -156,6 +253,9 @@ public final class Stage2Pipeline {
     private let ring = ReadyRing()
     private let presenter: MetalVideoPresenter
     private let decoder: VideoDecoder
+    /// Present cadence — `.arrival` (stage-2) or `.glass` (stage-3, the present gate). Fixed for
+    /// the pipeline's lifetime; SessionPresenter resolves it per session (see PresentPacing).
+    private let pacing: PresentPacing
     private let endToEndMeter: LatencyMeter?
     private let displayMeter: LatencyMeter?
     private let recovery = KeyframeRecovery()
@@ -190,14 +290,17 @@ public final class Stage2Pipeline {
     /// (received→decoded); `displayMeter` the display stage (decoded→on-glass, the ring wait +
     /// render + vsync — the tail stage-2 exists to shorten). All optional: metering never gates
     /// the presenter choice. Returns nil if Metal can't be set up (headless / no GPU) — caller
-    /// falls back to the stage-1 presenter.
+    /// falls back to the stage-1 presenter. `pacing` selects the stage-2 (arrival) vs stage-3
+    /// (glass-gated) present cadence — see PresentPacing.
     public init?(
         endToEndMeter: LatencyMeter?,
         decodeMeter: LatencyMeter? = nil,
-        displayMeter: LatencyMeter? = nil
+        displayMeter: LatencyMeter? = nil,
+        pacing: PresentPacing = .arrival
     ) {
         guard let presenter = MetalVideoPresenter.make() else { return nil }
         self.presenter = presenter
+        self.pacing = pacing
         self.endToEndMeter = endToEndMeter
         self.displayMeter = displayMeter
         let ring = ring
@@ -327,16 +430,29 @@ public final class Stage2Pipeline {
                 && UserDefaults.standard.bool(forKey: DefaultsKey.vsync))
         let debugStats = presentDebug ? PresentDebugStats() : nil
         let vsyncClock = vsyncClock
+        // Stage-3's one-in-flight present gate; nil = stage-2's present-on-arrival. A local (like
+        // the ring) so neither the render thread nor the presented handlers capture `self`.
+        let gate: PresentGate? = pacing == .glass ? PresentGate() : nil
         let renderThread = Thread {
             defer { renderStopped.signal() }
             while !token.isStopped {
                 if renderSignal.wait(timeout: .now() + .milliseconds(100)) == .timedOut {
-                    debugStats?.flushIfDue(ring: ring)
+                    debugStats?.flushIfDue(ring: ring, gate: gate)
+                    continue
+                }
+                // Stage-3: while a present is in flight, don't take from the ring at all — frames
+                // keep coalescing there (newest wins, the intended drop point) and the presented
+                // handler re-signals the moment the slot frees. Checked BEFORE the take so a gated
+                // frame is never bounced through putBack.
+                if let gate, !gate.tryAcquire(now: CACurrentMediaTime()) {
+                    debugStats?.gatedWake()
+                    debugStats?.flushIfDue(ring: ring, gate: gate)
                     continue
                 }
                 guard !token.isStopped, let frame = ring.take() else {
+                    gate?.release() // armed but nothing to render — don't hold the gate stale
                     debugStats?.emptyWake()
-                    debugStats?.flushIfDue(ring: ring)
+                    debugStats?.flushIfDue(ring: ring, gate: gate)
                     continue
                 }
                 // V-Sync ON: flip on the next predicted vsync (< one period out, stale link ⇒
@@ -347,6 +463,12 @@ public final class Stage2Pipeline {
                 let rendered = presenter.render(
                     frame.pixelBuffer, isHDR: frame.isHDR, presentAtMediaTime: presentAt
                 ) { presentedNs in
+                    // Stage-3: the flip reached glass (or was dropped) — free the present slot,
+                    // then re-signal so the freshest waiting ring frame goes out immediately.
+                    if let gate {
+                        gate.release()
+                        renderSignal.signal()
+                    }
                     // Fallback stamp for a dropped drawable (no system presentedTime): "now" on
                     // the Metal callback, converted to the CLOCK_REALTIME the meters live in.
                     let atNs = presentedNs
@@ -361,8 +483,11 @@ public final class Stage2Pipeline {
                 }
                 debugStats?.renderReturned(
                     ok: rendered, tookMs: (CACurrentMediaTime() - renderStarted) * 1000)
-                if !rendered { ring.putBack(frame) }
-                debugStats?.flushIfDue(ring: ring)
+                if !rendered {
+                    gate?.release() // no present registered — its handler will never fire
+                    ring.putBack(frame)
+                }
+                debugStats?.flushIfDue(ring: ring, gate: gate)
             }
         }
         renderThread.name = "punktfunk-stage2-render"

clients/apple/Sources/PunktfunkKit/Views/StreamView.swift

@@ -594,13 +594,12 @@ public final class StreamLayerView: NSView {
             guard let self, self.window?.isKeyWindow == true else { return }
             self.onDisconnectRequest?()
         }
-        capture.onToggleStats = { [weak self] in
+        capture.onCycleStats = { [weak self] in
             guard self?.window?.isKeyWindow == true else { return }
-            // Flip the shared setting directly — every @AppStorage reader (the HUD's visibility,
-            // the menu item's title) observes UserDefaults, so this is the same as the menu path.
-            let defaults = UserDefaults.standard
-            let current = defaults.object(forKey: DefaultsKey.hudEnabled) as? Bool ?? true
-            defaults.set(!current, forKey: DefaultsKey.hudEnabled)
+            // Advance the shared tier setting directly — every @AppStorage reader (the HUD's
+            // visibility/content, the Settings pickers) observes UserDefaults, so this is the
+            // same as the menu path.
+            StatsVerbosity.store(StatsVerbosity.current.next())
         }
         capture.start()
         inputCapture = capture

clients/apple/Tests/PunktfunkKitTests/PresentPacingTests.swift

@@ -0,0 +1,83 @@
+import XCTest
+
+#if canImport(Metal)
+import QuartzCore
+@testable import PunktfunkKit
+
+/// Stage-3 present pacing: the one-in-flight `PresentGate` and the stage-1/2/3 `PresenterChoice`
+/// resolution (setting + PUNKTFUNK_PRESENTER env override + the release-build stage-1 gate).
+final class PresentPacingTests: XCTestCase {
+    // MARK: - PresentGate
+
+    /// The core invariant: one present in flight. A second acquire while pending must fail (the
+    /// frame stays in the ring for the presented handler's re-signal); release reopens.
+    func testGateAdmitsOneInFlightPresent() {
+        let gate = PresentGate()
+        XCTAssertTrue(gate.tryAcquire(now: 0), "an idle gate must admit the first present")
+        XCTAssertFalse(gate.tryAcquire(now: 0.001), "a pending present must close the gate")
+        gate.release()
+        XCTAssertTrue(gate.tryAcquire(now: 0.002), "release must reopen the gate")
+        XCTAssertEqual(gate.drainForced(), 0, "no stale present was force-cleared")
+    }
+
+    /// The lost-handler insurance: a present whose handler never fires (the macOS "presents
+    /// aren't damage" hazard class) must not freeze the stream — past `staleAfter` the gate
+    /// force-opens and counts the event for the PUNKTFUNK_PRESENT_DEBUG `forced` stat.
+    func testGateForceOpensAfterStaleTimeout() {
+        let gate = PresentGate()
+        XCTAssertTrue(gate.tryAcquire(now: 10))
+        // Within the stale window the gate stays closed.
+        XCTAssertFalse(gate.tryAcquire(now: 10 + PresentGate.staleAfter - 0.01))
+        // Past it, the pending present is presumed lost: reopen, and count the force-clear.
+        XCTAssertTrue(gate.tryAcquire(now: 10 + PresentGate.staleAfter + 0.01))
+        XCTAssertEqual(gate.drainForced(), 1)
+        XCTAssertEqual(gate.drainForced(), 0, "drain resets the counter")
+    }
+
+    /// Release is idempotent (a late stale-path double-release must be harmless), and an
+    /// acquire-then-release with no present (empty ring after arming) leaves the gate clean.
+    func testGateReleaseIsIdempotent() {
+        let gate = PresentGate()
+        XCTAssertTrue(gate.tryAcquire(now: 0))
+        gate.release()
+        gate.release() // the stale-cleared present's handler firing late
+        XCTAssertTrue(gate.tryAcquire(now: 0.001))
+        XCTAssertEqual(gate.drainForced(), 0)
+    }
+
+    // MARK: - PresenterChoice
+
+    func testPresenterChoiceDefaultsToStage2() {
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: nil, env: nil, allowStage1: true), .stage2)
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "garbage", env: nil, allowStage1: true), .stage2)
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage2", env: nil, allowStage1: true), .stage2)
+    }
+
+    func testPresenterChoiceResolvesStage3FromSettingAndEnv() {
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage3", env: nil, allowStage1: true), .stage3)
+        // The env override wins over the persisted setting (A/B without touching settings)…
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage2", env: "stage3", allowStage1: true), .stage3)
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage3", env: "stage2", allowStage1: true), .stage2)
+        // …but an EMPTY env var is "unset", not an override.
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage3", env: "", allowStage1: true), .stage3)
+    }
+
+    /// Stage-1 (the freeze-prone system-layer diagnostic) resolves only where allowed (DEBUG
+    /// builds); a leftover "stage1" value in a release build maps back to stage-2.
+    func testPresenterChoiceGatesStage1() {
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage1", env: nil, allowStage1: true), .stage1)
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: "stage1", env: nil, allowStage1: false), .stage2)
+        XCTAssertEqual(
+            PresenterChoice.resolve(setting: nil, env: "stage1", allowStage1: false), .stage2)
+    }
+}
+#endif

clients/decky/README.md

@@ -76,6 +76,8 @@ restart is required for an out-of-band install to appear.
 | `src/page.tsx` | The `/punktfunk` fullscreen page — Hosts (with per-host details) / Settings / About tabs. |
 | `src/settings.tsx` · `src/pair.tsx` | Stream-settings section; the gamepad-navigable PIN-pairing modal. |
 | `src/library.tsx` | The per-host game picker (pin/unpin, "Open library on screen") + the pinned-game launch helper. |
+| `src/hostmgmt.tsx` | Add / edit host dialogs — mutate the shared known-hosts store (`client-known-hosts.json`) via the flatpak client's headless modes, so a host saved here shows up in the desktop client too. |
+| `src/ui.tsx` | Shared UI primitives for the fullscreen page + modals (right-aligned row actions, consistent Field layout). |
 | `src/hooks.ts` · `src/boundary.tsx` | Shared discovery/update/pins hooks + actions; the render error boundary. |
 | `src/steam.ts` | Steam-shortcut launch (`AddShortcut` / `SetAppLaunchOptions` / `RunGame`) — the focus-correct stream start. The shortcut's exe is `/bin/sh` with the wrapper passed as an argument, so the script never needs an exec bit (Decky's zip extraction drops it and the root-owned plugins dir can't be chmodded by the unprivileged backend). Launch extras ride env-prefix tokens: `PF_LAUNCH=<id>` (pinned game) / `PF_BROWSE=1` + `PF_MGMT=<port>` (on-screen library); ids are validated space/quote-free at pin AND launch time. |
 | `src/backend.ts` | Typed `callable` bridges to `main.py`. |

clients/linux/README.md

@@ -1,17 +1,20 @@
 # punktfunk — Linux client
 
 The native **Linux** app for streaming a punktfunk host to your desktop, laptop, or Steam Deck.
-It's a clean GTK4/libadwaita app that finds hosts on your network, pairs with a PIN, and puts a
-low-latency stream on glass at your display's own resolution and refresh rate.
+It's a clean relm4/GTK4/libadwaita **shell** that finds hosts on your network, pairs with a PIN,
+and manages your settings and library — the stream itself runs in the sibling
+**`punktfunk-session`** Vulkan binary ([`clients/session`](../session/README.md)), which the shell
+spawns, putting the picture on glass at your display's own resolution and refresh rate.
 
-Built in Rust, it links the shared **`punktfunk-core`** directly (no C ABI) and speaks the fast
+Built in Rust end to end (no C ABI): the shell shares its plumbing with the session binary through
+**`crates/pf-client-core`**, which links the **`punktfunk-core`** protocol crate and speaks the fast
 **`punktfunk/1`** protocol — QUIC control plane, GF(2¹⁶) FEC + AES-GCM data plane.
 
 ## Features
 
-- **Zero-copy hardware decode** — FFmpeg VAAPI decode → DRM-PRIME dmabuf → `GdkDmabufTexture`
-  (Tier-1 zero-copy on Intel and AMD), with an automatic software-HEVC fallback on NVIDIA or when
-  VAAPI is unavailable.
+- **Zero-copy hardware decode** — the session presenter decodes via **Vulkan Video** on every GPU
+  vendor (including NVIDIA), falling back to FFmpeg VAAPI → DRM-PRIME dmabuf and then software when
+  Vulkan Video is unavailable.
 - **Your display's native mode** — the host builds a virtual output at exactly your WxH@Hz; no
   scaling, no letterboxing. Steady 60 fps at 1080p60, ~6 ms capture→decoded on the LAN.
 - **Audio both ways** — PipeWire playback with a jitter ring, plus mic uplink to the host.
@@ -26,10 +29,10 @@ Built in Rust, it links the shared **`punktfunk-core`** directly (no C ABI) and
   shows its games (Steam + custom) as a poster grid; click one to launch it in the session.
   Fetched from the host's management API over mTLS — paired devices are authorized by their
   certificate, no extra host setup.
-- **Gamepad library launcher** (`--browse host`) — a console-style, controller-driven coverflow of
-  a paired host's library (drifting aurora backdrop, center-focus posters, button hints): A plays
-  the focused title, B quits, L1/R1 jump. Built for the Steam Deck plugin's "Open library" launch;
-  session end returns to the launcher. Arrow keys/Enter/Esc drive it too (no pad needed).
+- **Gamepad library launcher** (`--browse host`) — a console-style, controller-driven library view
+  of a paired host's games, rendered by the session binary's Skia console UI: A plays the focused
+  title, B quits, L1/R1 jump. Built for the Steam Deck plugin's "Open library" launch; session end
+  returns to the launcher. Arrow keys/Enter/Esc drive it too (no pad needed).
 
 ## Get it
 

clients/linux/src/app.rs

@@ -670,7 +670,7 @@ pub fn shortcuts_window(parent: &adw::ApplicationWindow) -> gtk::ShortcutsWindow
             </child>
             <child>
               <object class="GtkShortcutsShortcut">
-                <property name="title">Toggle statistics overlay</property>
+                <property name="title">Cycle the statistics overlay (off · compact · normal · detailed)</property>
                 <property name="accelerator">&lt;Control&gt;&lt;Alt&gt;&lt;Shift&gt;s</property>
               </object>
             </child>

clients/linux/src/ui_settings.rs

@@ -3,6 +3,7 @@
 
 use crate::trust::Settings;
 use adw::prelude::*;
+use pf_client_core::trust::StatsVerbosity;
 use std::cell::{Cell, RefCell};
 use std::rc::Rc;
 
@@ -324,10 +325,13 @@ pub fn show(
             "Software",
         ],
     );
-    let stats_row = adw::SwitchRow::builder()
-        .title("Show statistics overlay")
-        .subtitle("fps · bitrate · latency on the stream — Ctrl+Alt+Shift+S toggles live")
-        .build();
+    let stats_row = ChoiceRow::new(
+        &dialog,
+        inline,
+        "Statistics overlay",
+        "Compact = fps · latency · bitrate in one line — Ctrl+Alt+Shift+S cycles the tiers live",
+        &["Off", "Compact", "Normal", "Detailed"],
+    );
     let fullscreen_row = adw::SwitchRow::builder()
         .title("Start streams in fullscreen")
         .subtitle("F11, the mouse at the top edge, or L1+R1+Start+Select lead back out")
@@ -338,7 +342,7 @@ pub fn show(
     stream.add(compositor_row.widget());
     stream.add(decoder_row.widget());
     stream.add(&fullscreen_row);
-    stream.add(&stats_row);
+    stream.add(stats_row.widget());
 
     let input = adw::PreferencesGroup::builder().title("Input").build();
     // Which physical controller forwards as pad 0: automatic = the most recently connected
@@ -483,7 +487,11 @@ pub fn show(
         compositor_row.set_selected(comp_i as u32);
         let dec_i = DECODERS.iter().position(|&d| d == s.decoder).unwrap_or(0);
         decoder_row.set_selected(dec_i as u32);
-        stats_row.set_active(s.show_stats);
+        let stats_i = StatsVerbosity::ALL
+            .iter()
+            .position(|v| *v == s.stats_verbosity())
+            .unwrap_or(0);
+        stats_row.set_selected(stats_i as u32);
         fullscreen_row.set_active(s.fullscreen_on_stream);
         inhibit_row.set_active(s.inhibit_shortcuts);
         mic_row.set_active(s.mic_enabled);
@@ -509,7 +517,9 @@ pub fn show(
         s.compositor = COMPOSITORS[(compositor_row.selected() as usize).min(COMPOSITORS.len() - 1)]
             .to_string();
         s.decoder = DECODERS[(decoder_row.selected() as usize).min(DECODERS.len() - 1)].to_string();
-        s.show_stats = stats_row.is_active();
+        s.set_stats_verbosity(
+            StatsVerbosity::ALL[(stats_row.selected() as usize).min(StatsVerbosity::ALL.len() - 1)],
+        );
         s.fullscreen_on_stream = fullscreen_row.is_active();
         s.inhibit_shortcuts = inhibit_row.is_active();
         s.mic_enabled = mic_row.is_active();

clients/probe/src/main.rs

@@ -145,11 +145,35 @@ fn load_or_create_identity() -> Result<(String, String)> {
     let dir = std::path::PathBuf::from(home).join(".config/punktfunk");
     let (cp, kp) = (dir.join("client-cert.pem"), dir.join("client-key.pem"));
     if let (Ok(c), Ok(k)) = (std::fs::read_to_string(&cp), std::fs::read_to_string(&kp)) {
+        // Re-lock a store an older build left world-readable (this key is shared with the other
+        // clients' `~/.config/punktfunk/client-key.pem`); best-effort.
+        #[cfg(unix)]
+        {
+            use std::os::unix::fs::PermissionsExt;
+            let _ = std::fs::set_permissions(&dir, std::fs::Permissions::from_mode(0o700));
+            let _ = std::fs::set_permissions(&kp, std::fs::Permissions::from_mode(0o600));
+        }
         return Ok((c, k));
     }
     let (c, k) = endpoint::generate_identity().map_err(|e| anyhow!("generate identity: {e}"))?;
     std::fs::create_dir_all(&dir)?;
+    // The certificate is public; the key is the mTLS credential a paired host authorizes for full
+    // remote control, so it must not be world-readable — create it 0600 (a plain `fs::write`
+    // honors the umask → typically 0644).
     std::fs::write(&cp, &c)?;
+    #[cfg(unix)]
+    {
+        use std::os::unix::fs::{OpenOptionsExt, PermissionsExt};
+        let _ = std::fs::set_permissions(&dir, std::fs::Permissions::from_mode(0o700));
+        let mut f = std::fs::OpenOptions::new()
+            .write(true)
+            .create(true)
+            .truncate(true)
+            .mode(0o600)
+            .open(&kp)?;
+        f.write_all(k.as_bytes())?;
+    }
+    #[cfg(not(unix))]
     std::fs::write(&kp, &k)?;
     tracing::info!(cert = %cp.display(), "generated client identity");
     Ok((c, k))

clients/session/README.md

@@ -21,7 +21,8 @@ Reads the same identity / known-hosts / settings stores as the desktop client
 connects to a host it has no pinned fingerprint for (`--fp HEX` overrides the store).
 
 Stdout is the machine interface: `{"ready":true}` after the first presented frame,
-`stats: …` once per second (Ctrl+Alt+Shift+S toggles, `--stats` forces on), one
+`stats: …` once per second while the overlay tier isn't Off (always the full detailed
+text, whatever the OSD shows; `--stats` forces the overlay on), one
 `{"error"|"ended": …}` JSON line on the way out. Logs go to stderr. Exit codes: `0`
 clean end, `2` connect failed, `3` trust rejected / pairing required, `4` presenter
 init failed.
@@ -31,21 +32,24 @@ releases), Ctrl+Alt+Shift+D disconnects, F11 toggles fullscreen; the controller
 chord (L1+R1+Start+Select, hold to disconnect) works the same.
 
 The default build carries the Skia console UI (`ui` feature): the stats OSD and capture
-hint render in-window (Ctrl+Alt+Shift+S toggles both the OSD and the stdout mirror).
+hint render in-window. Ctrl+Alt+Shift+S cycles the OSD tier live — Off → Compact (one
+line: fps · latency · Mb/s) → Normal (mode + end-to-end percentiles) → Detailed (decoder
+path + per-stage latency equation); any tier but Off also emits the stdout mirror.
 `--no-default-features` is the ~5 MB power-user build — same streaming, stats on stdout
 only, no Skia anywhere in the dependency tree.
 
-Decode follows the Settings preference (auto: Vulkan Video → VAAPI → software):
-FFmpeg's Vulkan Video decoder runs on the presenter's own device where the stack
-supports it (every vendor, zero copy); VAAPI dmabufs import per-plane elsewhere;
-software is the universal fallback. 10-bit Main10 and HDR10 are advertised
+Decode follows the Settings preference (auto: Vulkan Video → VAAPI → software on Linux,
+Vulkan Video → D3D11VA → software on Windows): FFmpeg's Vulkan Video decoder runs on the
+presenter's own device where the stack supports it (every vendor, zero copy); VAAPI
+dmabufs import per-plane elsewhere (D3D11VA textures on Windows); software is the
+universal fallback. 10-bit Main10 and HDR10 are advertised
 (`VIDEO_CAP_10BIT|HDR`): P010 decodes through all three paths, and PQ streams present
 on an HDR10/ST.2084 swapchain when the desktop offers one (KDE HDR, gamescope) or
 tone-map in-shader to SDR when it doesn't (`PUNKTFUNK_TONEMAP_PEAK` tunes the rolloff,
 default ≈1000 nits). The host still gates the upgrade behind its `PUNKTFUNK_10BIT`
 policy.
 
-Debug/bisect knobs: `PUNKTFUNK_DECODER=vulkan|vaapi|software`, `PUNKTFUNK_PRESENT_MODE=
+Debug/bisect knobs: `PUNKTFUNK_DECODER=vulkan|vaapi|d3d11va|software`, `PUNKTFUNK_PRESENT_MODE=
 mailbox|immediate` (default FIFO), `PUNKTFUNK_VK_DEVICE=<index>` (multi-GPU), and
 `PUNKTFUNK_HW_FAULT=import` (fault every VAAPI dmabuf import — proves the three-strike
 demotion to software on healthy hardware).

clients/session/src/console.rs

@@ -135,7 +135,11 @@ pub fn run(target: Option<&str>) -> u8 {
         ),
         fullscreen: fullscreen_mode(),
         window_pos: window_pos(),
-        print_stats: settings_at_start.show_stats || arg_flag("--stats"),
+        // `--stats` forces the overlay visible without demoting a richer chosen tier.
+        stats_verbosity: match settings_at_start.stats_verbosity() {
+            trust::StatsVerbosity::Off if arg_flag("--stats") => trust::StatsVerbosity::Normal,
+            v => v,
+        },
         json_status,
         on_connected: Some(Box::new(|fingerprint: [u8; 32]| {
             trust::touch_last_used(&trust::hex(&fingerprint));

clients/session/src/main.rs

@@ -322,7 +322,12 @@ mod session_main {
             window_title: format!("Punktfunk · {title}"),
             fullscreen,
             window_pos: window_pos(),
-            print_stats: settings.show_stats || arg_flag("--stats"),
+            // `--stats` forces the overlay visible (tooling/debug runs) without
+            // demoting an explicitly chosen richer tier.
+            stats_verbosity: match settings.stats_verbosity() {
+                trust::StatsVerbosity::Off if arg_flag("--stats") => trust::StatsVerbosity::Normal,
+                v => v,
+            },
             json_status: true,
             on_connected: Some(Box::new(|fingerprint: [u8; 32]| {
                 // This host's card carries the accent bar in the desktop client now.

clients/windows/README.md

@@ -26,7 +26,8 @@ the fast **`punktfunk/1`** protocol.
   (probe burst over the real data plane → recommended bitrate, applied in one tap) and **forget**.
 - **Polished shell** — host cards, settings (resolution / refresh / host compositor / decoder /
   codec / bitrate / HDR / forwarded controller / gamepad type / system shortcuts / audio channels /
-  mic), a status-chip stream HUD, and the full trust surface. Stream input uses Win32 low-level
+  mic / stats-overlay level), the tiered stats overlay (Off / Compact / Normal / Detailed —
+  Ctrl+Alt+Shift+S cycles it live in the session window), and the full trust surface. Stream input uses Win32 low-level
   hooks with Moonlight-style capture: Ctrl+Alt+Shift+Q releases the pointer, a click on the stream
   re-captures it, and system shortcuts (Alt+Tab, Win, …) can act locally or forward to the host.
 

clients/windows/src/app/help.rs

@@ -19,7 +19,10 @@ const STREAM_SHORTCUTS: &[(&str, &str)] = &[
         "Release captured input (click the stream to recapture)",
     ),
     ("Ctrl+Alt+Shift+D", "Disconnect"),
-    ("Ctrl+Alt+Shift+S", "Toggle the statistics overlay"),
+    (
+        "Ctrl+Alt+Shift+S",
+        "Cycle the statistics overlay (off \u{00B7} compact \u{00B7} normal \u{00B7} detailed)",
+    ),
     (
         "LB+RB+Start+Back",
         "Controller: release input / leave fullscreen \u{2014} hold to disconnect",

clients/windows/src/app/settings.rs

@@ -4,6 +4,7 @@
 use super::style::*;
 use super::{AppCtx, Screen};
 use crate::trust::Settings;
+use pf_client_core::trust::StatsVerbosity;
 use punktfunk_core::config::GamepadPref;
 use std::sync::Arc;
 use windows_reactor::*;
@@ -46,6 +47,14 @@ const GAMEPADS: &[(&str, &str)] = &[
     ("xboxone", "Xbox One"),
     ("dualshock4", "DualShock 4"),
 ];
+/// Stats-overlay tiers: `(stored value, display label)` — the cross-client verbosity ladder
+/// (Compact ⊂ Normal ⊂ Detailed); Ctrl+Alt+Shift+S cycles it live in the session window.
+const STATS_TIERS: &[(StatsVerbosity, &str)] = &[
+    (StatsVerbosity::Off, "Off"),
+    (StatsVerbosity::Compact, "Compact"),
+    (StatsVerbosity::Normal, "Normal"),
+    (StatsVerbosity::Detailed, "Detailed"),
+];
 /// Host compositor presets: `(stored value, display label)`. Advisory — the host falls back to
 /// auto-detect when the choice is unavailable. Only meaningful against a Linux host.
 const COMPOSITORS: &[(&str, &str)] = &[
@@ -328,15 +337,14 @@ pub(crate) fn settings_page(
     )
     .tooltip("Sends the default microphone to the host's virtual mic source.");
 
-    let hud_toggle = setting_toggle(
-        ctx,
-        "Show the stats overlay (HUD)",
-        s.show_stats,
-        |s, on| s.show_stats = on,
-    )
+    let (hud_names, hud_i) = presets(STATS_TIERS, |v| *v == s.stats_verbosity());
+    let hud_combo = setting_combo(ctx, "Stats overlay (HUD)", hud_names, hud_i, |s, i| {
+        s.set_stats_verbosity(STATS_TIERS[i].0);
+    })
     .tooltip(
-        "The in-stream overlay: mode, codec, fps, bitrate, latency, decode path. \
-         Ctrl+Alt+Shift+S toggles it live while streaming.",
+        "How much the in-stream overlay shows: Compact (fps \u{00B7} latency \u{00B7} bitrate \
+         in one line) \u{2192} Normal \u{2192} Detailed (decode path and per-stage latency). \
+         Ctrl+Alt+Shift+S cycles the tiers live while streaming.",
     );
 
     let licenses_button = {
@@ -368,7 +376,7 @@ pub(crate) fn settings_page(
                     codec_combo.into(),
                     bitrate_box.into(),
                     hdr_toggle.into(),
-                    hud_toggle.into(),
+                    hud_combo.into(),
                 ]);
                 controls
             }),

crates/pf-client-core/src/trust.rs

@@ -34,16 +34,61 @@ pub fn load_or_create_identity() -> Result<(String, String)> {
     let dir = config_dir()?;
     let (cp, kp) = (dir.join("client-cert.pem"), dir.join("client-key.pem"));
     if let (Ok(c), Ok(k)) = (std::fs::read_to_string(&cp), std::fs::read_to_string(&kp)) {
+        // An older build wrote the key with a plain `fs::write`, which honors the umask and
+        // typically lands 0644 — world-readable. Re-lock an existing store on load so upgrades
+        // get fixed, not just fresh installs. Best-effort (a read-only store keeps what it has).
+        #[cfg(unix)]
+        lock_identity_perms(&dir, &kp);
         return Ok((c, k));
     }
     let (c, k) = endpoint::generate_identity().map_err(|e| anyhow!("generate identity: {e}"))?;
     std::fs::create_dir_all(&dir)?;
+    // The private key authorizes this client for full remote control of a paired host, so it must
+    // never be world-readable: lock the dir to the owner (0700) and create the key 0600 from the
+    // start (`fs::write` alone honors the umask → typically 0644). The certificate is public. On
+    // non-Unix the %APPDATA% profile ACL already scopes the dir to the user, so std perms suffice.
+    #[cfg(unix)]
+    {
+        use std::os::unix::fs::PermissionsExt;
+        std::fs::set_permissions(&dir, std::fs::Permissions::from_mode(0o700))?;
+    }
     std::fs::write(&cp, &c)?;
-    std::fs::write(&kp, &k)?;
+    write_private_key(&kp, k.as_bytes())?;
     tracing::info!(cert = %cp.display(), "generated client identity");
     Ok((c, k))
 }
 
+/// Write the client's mTLS private key owner-only. On Unix the file is created with mode 0600 from
+/// the outset — an `fs::write` + later `chmod` would briefly expose it at the umask default. On
+/// other platforms std's default perms plus the %APPDATA% profile ACL scope it to the user.
+fn write_private_key(path: &std::path::Path, bytes: &[u8]) -> Result<()> {
+    #[cfg(unix)]
+    {
+        use std::io::Write;
+        use std::os::unix::fs::OpenOptionsExt;
+        let mut f = std::fs::OpenOptions::new()
+            .write(true)
+            .create(true)
+            .truncate(true)
+            .mode(0o600)
+            .open(path)?;
+        f.write_all(bytes)?;
+    }
+    #[cfg(not(unix))]
+    std::fs::write(path, bytes)?;
+    Ok(())
+}
+
+/// Best-effort re-lock of an already-present identity (dir 0700, key 0600) — for stores written by
+/// an older build that left the key world-readable. Errors are ignored: the worst case is the
+/// pre-existing perms, which this never loosens.
+#[cfg(unix)]
+fn lock_identity_perms(dir: &std::path::Path, key: &std::path::Path) {
+    use std::os::unix::fs::PermissionsExt;
+    let _ = std::fs::set_permissions(dir, std::fs::Permissions::from_mode(0o700));
+    let _ = std::fs::set_permissions(key, std::fs::Permissions::from_mode(0o600));
+}
+
 pub fn hex(fp: &[u8; 32]) -> String {
     fp.iter().map(|b| format!("{b:02x}")).collect()
 }
@@ -254,6 +299,50 @@ pub fn probe_reachable_many(
         .collect()
 }
 
+/// How much the on-stream statistics overlay shows — the Android client's tiers, shared
+/// across every client (design/stats-unification.md): each tier is a strict superset of
+/// the previous. Ctrl+Alt+Shift+S cycles Off → Compact → Normal → Detailed live.
+#[derive(Clone, Copy, PartialEq, Eq, Debug, Serialize, Deserialize)]
+#[serde(rename_all = "lowercase")]
+pub enum StatsVerbosity {
+    Off,
+    /// One glanceable line: fps · end-to-end ms · Mb/s.
+    Compact,
+    /// Stream mode plus the end-to-end latency percentiles and loss counters.
+    Normal,
+    /// Everything: decoder path, HDR tags, and the per-stage latency equation.
+    Detailed,
+}
+
+impl StatsVerbosity {
+    /// Cycle order (also the settings pickers' option order).
+    pub const ALL: [StatsVerbosity; 4] = [
+        StatsVerbosity::Off,
+        StatsVerbosity::Compact,
+        StatsVerbosity::Normal,
+        StatsVerbosity::Detailed,
+    ];
+
+    /// The next tier in the live cycle, wrapping back to Off.
+    pub fn next(self) -> StatsVerbosity {
+        match self {
+            StatsVerbosity::Off => StatsVerbosity::Compact,
+            StatsVerbosity::Compact => StatsVerbosity::Normal,
+            StatsVerbosity::Normal => StatsVerbosity::Detailed,
+            StatsVerbosity::Detailed => StatsVerbosity::Off,
+        }
+    }
+
+    pub fn label(self) -> &'static str {
+        match self {
+            StatsVerbosity::Off => "Off",
+            StatsVerbosity::Compact => "Compact",
+            StatsVerbosity::Normal => "Normal",
+            StatsVerbosity::Detailed => "Detailed",
+        }
+    }
+}
+
 /// App settings, persisted as JSON. Stringly-typed gamepad/compositor prefs so the file
 /// stays readable; parsed with `*Pref::from_name` at connect time.
 #[derive(Clone, Serialize, Deserialize)]
@@ -301,10 +390,16 @@ pub struct Settings {
     /// `default = true`: the Linux stores never carried this and always advertised.
     #[serde(default = "default_true")]
     pub hdr_enabled: bool,
-    /// Show the on-stream statistics overlay (toggle live with Ctrl+Alt+Shift+S).
-    /// `alias`: the pre-unification WinUI shell (≤ 0.8.4) persisted this as `show_hud`.
+    /// Legacy on/off for the stats overlay — superseded by `stats_verbosity` but kept
+    /// written in sync (`set_stats_verbosity`) so pre-tier binaries reading the same
+    /// file keep working. `alias`: the pre-unification WinUI shell (≤ 0.8.4) persisted
+    /// this as `show_hud`.
     #[serde(alias = "show_hud")]
     pub show_stats: bool,
+    /// Stats overlay tier. `None` = a pre-tier store; resolve through
+    /// [`Settings::stats_verbosity`], which falls back to `show_stats`.
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub stats_verbosity: Option<StatsVerbosity>,
     /// Enter fullscreen when a stream starts (F11 / the controller chord / the top-edge
     /// header reveal exit it). Gaming-Mode launches (`--fullscreen`) fullscreen regardless.
     pub fullscreen_on_stream: bool,
@@ -322,6 +417,23 @@ fn default_true() -> bool {
 }
 
 impl Settings {
+    /// The stats-overlay tier, resolving pre-tier stores: an old `show_stats = false`
+    /// reads as Off, everything else as Normal (≈ what the pre-tier overlay showed).
+    pub fn stats_verbosity(&self) -> StatsVerbosity {
+        self.stats_verbosity.unwrap_or(if self.show_stats {
+            StatsVerbosity::Normal
+        } else {
+            StatsVerbosity::Off
+        })
+    }
+
+    /// Set the tier, keeping the legacy `show_stats` bool coherent for pre-tier
+    /// binaries that read the same settings file.
+    pub fn set_stats_verbosity(&mut self, v: StatsVerbosity) {
+        self.stats_verbosity = Some(v);
+        self.show_stats = v != StatsVerbosity::Off;
+    }
+
     /// The `codec` setting as a `quic::CODEC_*` preference bit (`0` = auto).
     pub fn preferred_codec(&self) -> u8 {
         match self.codec.as_str() {
@@ -351,6 +463,7 @@ impl Default for Settings {
             adapter: String::new(),
             hdr_enabled: true,
             show_stats: true,
+            stats_verbosity: None,
             fullscreen_on_stream: true,
             library_enabled: false,
         }
@@ -432,6 +545,28 @@ mod tests {
         assert!(!s.library_enabled);
     }
 
+    /// Stats-tier resolution: a pre-tier store falls back to `show_stats` (off → Off,
+    /// on/absent → Normal), an explicit tier wins, and setting a tier keeps the legacy
+    /// bool in sync so pre-tier binaries reading the same file agree on off vs on.
+    #[test]
+    fn stats_verbosity_migrates_and_round_trips() {
+        let mut s: Settings = serde_json::from_str("{}").unwrap();
+        assert_eq!(s.stats_verbosity(), StatsVerbosity::Normal);
+        let off: Settings = serde_json::from_str(r#"{"show_stats":false}"#).unwrap();
+        assert_eq!(off.stats_verbosity(), StatsVerbosity::Off);
+
+        s.set_stats_verbosity(StatsVerbosity::Compact);
+        assert!(s.show_stats);
+        s.set_stats_verbosity(StatsVerbosity::Off);
+        assert!(!s.show_stats);
+
+        s.set_stats_verbosity(StatsVerbosity::Detailed);
+        let round: Settings = serde_json::from_str(&serde_json::to_string(&s).unwrap()).unwrap();
+        assert_eq!(round.stats_verbosity(), StatsVerbosity::Detailed);
+        // The tier serializes lowercase — the file stays human-readable.
+        assert!(serde_json::to_string(&s).unwrap().contains("\"detailed\""));
+    }
+
     /// The WinUI shell's known-hosts shape (no `last_used` field) loads losslessly — same
     /// filename, same directory, so on Windows the two clients genuinely share the store.
     #[test]

crates/pf-console-ui/src/screens/settings.rs

@@ -10,6 +10,7 @@ use crate::screens::{Ctx, Outbox};
 use crate::theme::{Fonts, DIM, W};
 use crate::widgets::{ListMsg, MenuList, RowSpec};
 use pf_client_core::gamepad::{MenuEvent, MenuPulse};
+use pf_client_core::trust::StatsVerbosity;
 use skia_safe::{Canvas, Rect};
 
 /// Stable row identity — adjust/activate dispatch by id so nothing acts on a stale
@@ -241,7 +242,7 @@ fn row_spec(id: RowId, ctx: &Ctx) -> RowSpec {
         RowId::Stats => (
             Some("Interface"),
             "Statistics overlay",
-            on_off(s.show_stats).into(),
+            s.stats_verbosity().label().into(),
         ),
     };
     RowSpec {
@@ -274,7 +275,10 @@ fn detail(id: RowId) -> &'static str {
         RowId::Mic => "Send this device's microphone to the host's virtual mic.",
         RowId::Pad => "Which pad is forwarded to the host, as player 1.",
         RowId::PadType => "The virtual pad the host creates — Automatic matches this controller.",
-        RowId::Stats => "Resolution, frame rate, throughput and latency while streaming.",
+        RowId::Stats => {
+            "How much the overlay shows: Compact (one line) → Normal → Detailed. \
+             Ctrl+Alt+Shift+S cycles it live while streaming."
+        }
     }
 }
 
@@ -332,7 +336,13 @@ fn adjust(id: RowId, delta: i32, wrap: bool, ctx: &mut Ctx) -> bool {
             step_option(cur, keys.len(), delta, wrap).map(|i| s.forward_pad = keys[i].clone())
         }
         RowId::PadType => step_str(&PAD_TYPES, &mut s.gamepad, delta, wrap),
-        RowId::Stats => toggle(&mut s.show_stats, delta, wrap),
+        RowId::Stats => {
+            let cur = StatsVerbosity::ALL
+                .iter()
+                .position(|v| *v == s.stats_verbosity());
+            step_option(cur, StatsVerbosity::ALL.len(), delta, wrap)
+                .map(|i| s.set_stats_verbosity(StatsVerbosity::ALL[i]))
+        }
     }
     .is_some()
 }

crates/pf-driver-proto/README.md

@@ -13,4 +13,4 @@ Defining every wire struct here — with `const` size/offset asserts and `bytemu
 host↔driver ABI drift into a **compile error** instead of a silent frame or IOCTL corruption.
 
 See the crate root ([`src/`](src/)) for the wire types; the Windows virtual-display design is in
-[`design/windows-virtual-display-rust-port.md`](../../design/windows-virtual-display-rust-port.md).
+the internal planning repo (punktfunk-planning: `windows-virtual-display-rust-port.md`).

crates/pf-presenter/src/run.rs

@@ -8,8 +8,10 @@
 //! library; the app quits only on B/window-close).
 //!
 //! Stdout is the machine interface (the shell↔session contract): one `{"ready":true}`
-//! line after the first presented frame, `stats: …` lines once per window while enabled
-//! (Ctrl+Alt+Shift+S toggles). Logs go to stderr (the binary configures tracing so).
+//! line after the first presented frame, `stats: …` lines once per window while the
+//! overlay tier isn't Off (Ctrl+Alt+Shift+S cycles Off → Compact → Normal → Detailed;
+//! the stdout line always carries the full Detailed text so parsers see a stable
+//! shape). Logs go to stderr (the binary configures tracing so).
 
 use crate::input::Capture;
 use crate::overlay::{FrameCtx, Overlay, OverlayAction, OverlayFrame, SessionPhase};
@@ -17,6 +19,7 @@ use crate::vk::{FrameInput, Presenter};
 use anyhow::{Context as _, Result};
 use pf_client_core::gamepad::GamepadService;
 use pf_client_core::session::{self, SessionEvent, SessionHandle, SessionParams, Stats};
+use pf_client_core::trust::StatsVerbosity;
 use pf_client_core::video::VulkanDecodeDevice;
 use pf_client_core::video::{DecodedFrame, DecodedImage};
 use punktfunk_core::client::NativeClient;
@@ -37,8 +40,9 @@ pub struct SessionOpts {
     /// changing content, not a window jumping displays). Fullscreen follows the display
     /// this lands on.
     pub window_pos: Option<(i32, i32)>,
-    /// Print `stats:` lines (Ctrl+Alt+Shift+S toggles live).
-    pub print_stats: bool,
+    /// Stats overlay tier at start — gates the OSD panel AND the stdout `stats:` lines
+    /// (Ctrl+Alt+Shift+S cycles Off → Compact → Normal → Detailed live).
+    pub stats_verbosity: StatsVerbosity,
     /// Emit the `{"ready":true}` stdout line after the first presented frame.
     pub json_status: bool,
     /// Called once on `Connected` with the host's fingerprint (trust persistence is the
@@ -162,8 +166,11 @@ struct StreamState {
     // all) demotes the decoder to software via the shared flag — once per session.
     dmabuf_demoted: bool,
     hw_fails: u32,
-    /// The OSD's text (multi-line; rebuilt each Stats window).
+    /// The OSD's text (multi-line; rebuilt each Stats window and on a live tier cycle).
     osd_text: String,
+    /// The last pump window, kept so a Ctrl+Alt+Shift+S tier cycle can re-render the
+    /// OSD immediately instead of waiting up to 1 s for the next Stats event.
+    last_stats: Option<Stats>,
 }
 
 impl StreamState {
@@ -207,6 +214,7 @@ impl StreamState {
             dmabuf_demoted: false,
             hw_fails: 0,
             osd_text: String::new(),
+            last_stats: None,
         }
     }
 
@@ -351,7 +359,7 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
     let mouse = sdl.mouse();
 
     let mut fullscreen = opts.fullscreen;
-    let mut print_stats = opts.print_stats;
+    let mut stats_verbosity = opts.stats_verbosity;
     let mut overlay_frame: Option<OverlayFrame> = None;
     // SDL text input tracks the overlay's editing state (started = IME/`TextInput`
     // events on desktop, and the door Steam's on-screen keyboard types through under
@@ -452,7 +460,24 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
                         continue;
                     }
                     if chord && sc == Scancode::S {
-                        print_stats = !print_stats;
+                        stats_verbosity = stats_verbosity.next();
+                        tracing::info!(tier = ?stats_verbosity, "chord: stats verbosity");
+                        // Re-render the OSD from the last window immediately — waiting
+                        // for the next Stats event would lag the keypress by up to 1 s.
+                        if let Some(st) = &mut stream {
+                            let text = match &st.last_stats {
+                                Some(s) => stats_text(
+                                    stats_verbosity,
+                                    &st.mode_line,
+                                    s,
+                                    &st.presented,
+                                    st.hdr,
+                                    presenter.hdr_active(),
+                                ),
+                                None => String::new(),
+                            };
+                            st.osd_text = text;
+                        }
                         continue;
                     }
                     // F11 or Alt+Enter (some keyboards' Fn layer sends a media key for
@@ -647,15 +672,27 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
                 }
                 SessionEvent::Stats(s) => {
                     st.osd_text = stats_text(
+                        stats_verbosity,
                         &st.mode_line,
                         &s,
                         &st.presented,
                         st.hdr,
                         presenter.hdr_active(),
                     );
-                    if print_stats {
-                        println!("stats: {}", st.osd_text.replace('\n', " | "));
+                    if stats_verbosity != StatsVerbosity::Off {
+                        // The stdout line is the machine interface (shell status card,
+                        // scripts) — always the full Detailed text, whatever the OSD tier.
+                        let full = stats_text(
+                            StatsVerbosity::Detailed,
+                            &st.mode_line,
+                            &s,
+                            &st.presented,
+                            st.hdr,
+                            presenter.hdr_active(),
+                        );
+                        println!("stats: {}", full.replace('\n', " | "));
                     }
+                    st.last_stats = Some(s);
                 }
                 SessionEvent::Failed {
                     msg,
@@ -728,7 +765,8 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
                         _ => None,
                     };
                     (
-                        (print_stats && !st.osd_text.is_empty()).then_some(st.osd_text.as_str()),
+                        (stats_verbosity != StatsVerbosity::Off && !st.osd_text.is_empty())
+                            .then_some(st.osd_text.as_str()),
                         hint,
                     )
                 }
@@ -996,45 +1034,138 @@ const HINT_KEYBOARD: &str = "Click the stream to capture input · Ctrl+Alt+Shift
 const HINT_WITH_PAD: &str = "Click the stream to capture input · Ctrl+Alt+Shift+Q releases · \
      Ctrl+Alt+Shift+D disconnects · hold L1 + R1 + Start + Select to leave";
 
-/// The unified stats window (design/stats-unification.md) as OSD text — multi-line for
-/// the console-UI panel; the stdout `stats:` line joins it with `|`.
+/// The unified stats window (design/stats-unification.md) as OSD text at the given tier
+/// (the Android client's vocabulary, each a strict superset of the previous):
+/// Compact = one glanceable line, Normal = mode + end-to-end percentiles + loss,
+/// Detailed = decoder path, HDR tag and the per-stage equation on top. Off reads empty.
+/// Multi-line for the console-UI panel; the stdout `stats:` line joins Detailed with `|`.
 ///
 /// The HDR tag is honest about the display path: `HDR` only when the swapchain actually
 /// runs HDR10 (`hdr_display`); a PQ stream tone-mapped onto an SDR surface (no HDR10
 /// format offered, HDR off in the compositor) shows `HDR→SDR` instead.
 fn stats_text(
+    verbosity: StatsVerbosity,
     mode_line: &str,
     s: &Stats,
     p: &PresentedWindow,
     hdr_stream: bool,
     hdr_display: bool,
 ) -> String {
-    let mut text = format!(
-        "{mode_line} · {:.0} fps · {:.1} Mb/s · {}{}",
-        s.fps,
-        s.mbps,
-        if s.decoder.is_empty() { "-" } else { s.decoder },
-        match (hdr_stream, hdr_display) {
-            (true, true) => " · HDR",
-            (true, false) => " · HDR→SDR",
-            _ => "",
-        },
-    );
+    match verbosity {
+        StatsVerbosity::Off => return String::new(),
+        StatsVerbosity::Compact => {
+            // fps · e2e ms · Mb/s — the latency term waits for the first presenter
+            // window (0 = no capture→displayed samples yet).
+            let mut text = format!("{:.0} fps", s.fps);
+            if p.e2e_p50_ms > 0.0 {
+                text.push_str(&format!(" · {:.1} ms", p.e2e_p50_ms));
+            }
+            text.push_str(&format!(" · {:.0} Mb/s", s.mbps));
+            if s.lost > 0 {
+                text.push_str(&format!(" · lost {}", s.lost));
+            }
+            return text;
+        }
+        StatsVerbosity::Normal | StatsVerbosity::Detailed => {}
+    }
+    let detailed = verbosity == StatsVerbosity::Detailed;
+    let mut text = if detailed {
+        format!(
+            "{mode_line} · {:.0} fps · {:.1} Mb/s · {}{}",
+            s.fps,
+            s.mbps,
+            if s.decoder.is_empty() { "-" } else { s.decoder },
+            match (hdr_stream, hdr_display) {
+                (true, true) => " · HDR",
+                (true, false) => " · HDR→SDR",
+                _ => "",
+            },
+        )
+    } else {
+        format!("{mode_line} · {:.0} fps · {:.1} Mb/s", s.fps, s.mbps)
+    };
     text.push_str(&format!(
         "\ne2e {:.1}/{:.1} ms (p50/p95)",
         p.e2e_p50_ms, p.e2e_p95_ms
     ));
-    if s.split {
-        text.push_str(&format!(" · host {:.1} · net {:.1}", s.host_ms, s.net_ms));
-    } else {
-        text.push_str(&format!(" · host+net {:.1}", s.host_net_ms));
+    if detailed {
+        if s.split {
+            text.push_str(&format!(" · host {:.1} · net {:.1}", s.host_ms, s.net_ms));
+        } else {
+            text.push_str(&format!(" · host+net {:.1}", s.host_net_ms));
+        }
+        text.push_str(&format!(
+            " · decode {:.1} · display {:.1} ms",
+            s.decode_ms, p.display_ms
+        ));
     }
-    text.push_str(&format!(
-        " · decode {:.1} · display {:.1} ms",
-        s.decode_ms, p.display_ms
-    ));
     if s.lost > 0 {
         text.push_str(&format!("\nlost {} ({:.1}%)", s.lost, s.lost_pct));
     }
     text
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn sample() -> (Stats, PresentedWindow) {
+        (
+            Stats {
+                fps: 119.6,
+                mbps: 24.3,
+                host_net_ms: 2.1,
+                host_ms: 1.2,
+                net_ms: 0.9,
+                split: true,
+                decode_ms: 1.8,
+                lost: 3,
+                lost_pct: 0.4,
+                decoder: "vulkan",
+            },
+            PresentedWindow {
+                e2e_p50_ms: 6.4,
+                e2e_p95_ms: 9.1,
+                display_ms: 1.1,
+            },
+        )
+    }
+
+    /// The tier ladder: Off is empty, Compact is one line, Normal adds the mode + e2e
+    /// lines but no stage terms or decoder tag, Detailed carries everything.
+    #[test]
+    fn stats_text_tiers() {
+        let (s, p) = sample();
+        let text = |v| stats_text(v, "1920×1080@120", &s, &p, true, false);
+
+        assert_eq!(text(StatsVerbosity::Off), "");
+
+        let compact = text(StatsVerbosity::Compact);
+        assert_eq!(compact, "120 fps · 6.4 ms · 24 Mb/s · lost 3");
+        assert_eq!(compact.lines().count(), 1);
+
+        let normal = text(StatsVerbosity::Normal);
+        assert!(normal.starts_with("1920×1080@120 · 120 fps · 24.3 Mb/s\n"));
+        assert!(normal.contains("e2e 6.4/9.1 ms (p50/p95)"));
+        assert!(normal.contains("lost 3 (0.4%)"));
+        assert!(!normal.contains("vulkan"), "decoder tag is Detailed-only");
+        assert!(!normal.contains("decode"), "stage terms are Detailed-only");
+
+        let detailed = text(StatsVerbosity::Detailed);
+        assert!(detailed.contains("vulkan · HDR→SDR"));
+        assert!(detailed.contains("host 1.2 · net 0.9 · decode 1.8 · display 1.1 ms"));
+        assert!(detailed.contains("lost 3 (0.4%)"));
+    }
+
+    /// Compact omits the latency term until the presenter's first e2e window lands.
+    #[test]
+    fn compact_waits_for_e2e() {
+        let (mut s, _) = sample();
+        s.lost = 0;
+        let p = PresentedWindow::default();
+        assert_eq!(
+            stats_text(StatsVerbosity::Compact, "m", &s, &p, false, false),
+            "120 fps · 24 Mb/s"
+        );
+    }
+}

crates/punktfunk-core/README.md

@@ -57,5 +57,5 @@ bash crates/punktfunk-core/tests/c/run.sh    # standalone C-ABI link + round-tri
 
 - **[`punktfunk-host`](../punktfunk-host/README.md)** — the streaming host built on this core
 - **[Clients](../../clients/)** — the apps that link this core over the C ABI (or directly, in Rust)
-- **[`design/implementation-plan.md`](../../design/implementation-plan.md)** — why GF(2¹⁶) FEC, the
+- **punktfunk-planning: `implementation-plan.md`** (internal planning repo) — why GF(2¹⁶) FEC, the
   latency budget, and the architecture thesis

crates/punktfunk-core/src/client.rs

@@ -1242,6 +1242,7 @@ async fn worker_main(args: WorkerArgs) {
             welcome.chroma_format,
             welcome.audio_channels,
             welcome.codec,
+            welcome.host_caps,
         ))
     };
 
@@ -1261,6 +1262,7 @@ async fn worker_main(args: WorkerArgs) {
         chroma_format,
         audio_channels,
         codec,
+        host_caps,
     ) = match setup.await {
         Ok(t) => t,
         Err(e) => {
@@ -1282,11 +1284,55 @@ async fn worker_main(args: WorkerArgs) {
         codec,
     )));
 
-    // Input task: embedder events → QUIC datagrams.
+    // Input task: embedder events → QUIC datagrams. Toward a host that advertised
+    // HOST_CAP_GAMEPAD_STATE, the per-transition gamepad events every embedder still emits are
+    // folded into idempotent, sequence-numbered full-state snapshots (`GamepadSnapshot`): the
+    // datagram plane drops and reorders (and sheds oldest-first at the 4 KiB send cap), so a lost
+    // per-transition event would corrupt held pad state until the *next* change — a held trigger
+    // stuck wrong indefinitely. Snapshots heal on the next send, the seq lets the host drop stale
+    // reorders, and a periodic refresh of every touched pad bounds any loss to one refresh
+    // interval — the same idempotent-state discipline as the host's 500 ms rumble refresh.
+    // Keyboard/mouse/touch events pass through unchanged; an older host (no caps bit) keeps
+    // getting the legacy per-transition gamepad events.
     let input_conn = conn.clone();
+    let gamepad_snapshots = host_caps & crate::quic::HOST_CAP_GAMEPAD_STATE != 0;
     tokio::spawn(async move {
-        while let Some(ev) = input_rx.recv().await {
-            let _ = input_conn.send_datagram(ev.encode().to_vec().into());
+        use crate::input::{GamepadSnapshot, InputKind, MAX_PADS};
+        // Touched pads only: an entry appears on the first gamepad event for that index, so the
+        // refresh never conjures a virtual pad the embedder didn't drive.
+        let mut pads: [Option<GamepadSnapshot>; MAX_PADS] = [None; MAX_PADS];
+        let mut refresh = tokio::time::interval(Duration::from_millis(100));
+        refresh.set_missed_tick_behavior(tokio::time::MissedTickBehavior::Delay);
+        loop {
+            tokio::select! {
+                ev = input_rx.recv() => {
+                    let Some(ev) = ev else { break };
+                    let idx = ev.flags as usize;
+                    if gamepad_snapshots
+                        && matches!(ev.kind, InputKind::GamepadButton | InputKind::GamepadAxis)
+                        && idx < MAX_PADS
+                    {
+                        let snap = pads[idx].get_or_insert(GamepadSnapshot {
+                            pad: idx as u8,
+                            ..Default::default()
+                        });
+                        // Unknown axis ids don't send (the host's legacy fold drops them too).
+                        if snap.fold(&ev) {
+                            snap.seq = snap.seq.wrapping_add(1);
+                            let _ = input_conn
+                                .send_datagram(snap.to_event().encode().to_vec().into());
+                        }
+                        continue;
+                    }
+                    let _ = input_conn.send_datagram(ev.encode().to_vec().into());
+                }
+                _ = refresh.tick() => {
+                    for snap in pads.iter_mut().flatten() {
+                        snap.seq = snap.seq.wrapping_add(1);
+                        let _ = input_conn.send_datagram(snap.to_event().encode().to_vec().into());
+                    }
+                }
+            }
         }
     });
 

crates/punktfunk-core/src/input.rs

@@ -40,6 +40,17 @@ pub enum InputKind {
     TouchMove = 10,
     /// Touch ends. Only `code` (the touch id) is used.
     TouchUp = 11,
+    /// Full gamepad state in one event ([`GamepadSnapshot`]) — idempotent, sequence-numbered.
+    ///
+    /// The per-transition [`GamepadButton`](Self::GamepadButton)/[`GamepadAxis`](Self::GamepadAxis)
+    /// events are fragile on the unreliable datagram plane: a dropped or reordered event corrupts
+    /// the host's accumulated pad state until the *next* change (a held trigger stays wrong
+    /// indefinitely). A snapshot carries the whole pad, so loss heals on the next send and the
+    /// sequence number lets the host drop stale reorders — the same idempotent-state discipline
+    /// as the host→client rumble refresh. Sent only when the host advertised
+    /// [`HOST_CAP_GAMEPAD_STATE`](crate::quic::HOST_CAP_GAMEPAD_STATE); older hosts keep
+    /// receiving the per-transition events.
+    GamepadState = 12,
 }
 
 /// The gamepad wire contract for [`InputKind::GamepadButton`]/[`InputKind::GamepadAxis`].
@@ -111,11 +122,122 @@ impl InputKind {
             9 => TouchDown,
             10 => TouchMove,
             11 => TouchUp,
+            12 => GamepadState,
             _ => return None,
         })
     }
 }
 
+/// The number of gamepads addressable on the wire (`flags` pad index 0..15). Shared by the
+/// client's snapshot fold and the host's per-pad accumulators.
+pub const MAX_PADS: usize = 16;
+
+/// One pad's complete state, packed into a single [`InputKind::GamepadState`] event — the
+/// whole 18-byte wire layout is reused, nothing is appended:
+///
+/// - `code`  = `buttons` (the [`gamepad`] `BTN_*` bitmask, extended bits included)
+/// - `x`     = `ls_x << 16 | ls_y` (two i16 halves, wire stick convention: **+y = up**)
+/// - `y`     = `rs_x << 16 | rs_y`
+/// - `flags` = `seq << 24 | left_trigger << 16 | right_trigger << 8 | pad`
+///
+/// `seq` is a per-pad wrapping u8, bumped on every send (changes *and* refreshes); the host
+/// applies a snapshot only when `seq` is newer than the last applied one (wrapping i8
+/// compare), so a datagram the network reordered can't roll held state backwards. The wrap
+/// window (128 sends) dwarfs any real reorder window, and the client's periodic refresh
+/// heals the pathological case anyway.
+#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
+pub struct GamepadSnapshot {
+    /// Pad index 0..[`MAX_PADS`].
+    pub pad: u8,
+    /// Wrapping send counter (see the type docs for the reorder gate).
+    pub seq: u8,
+    /// [`gamepad`] `BTN_*` bitmask.
+    pub buttons: u32,
+    /// Triggers 0..255 (the [`gamepad::AXIS_LT`]/[`gamepad::AXIS_RT`] convention).
+    pub left_trigger: u8,
+    pub right_trigger: u8,
+    /// Sticks −32768..32767, **+y = up** (the wire convention).
+    pub ls_x: i16,
+    pub ls_y: i16,
+    pub rs_x: i16,
+    pub rs_y: i16,
+}
+
+impl GamepadSnapshot {
+    /// Pack into the fixed [`InputEvent`] layout (kind = [`InputKind::GamepadState`]).
+    pub fn to_event(&self) -> InputEvent {
+        InputEvent {
+            kind: InputKind::GamepadState,
+            _pad: [0; 3],
+            code: self.buttons,
+            x: ((self.ls_x as u16 as i32) << 16) | (self.ls_y as u16 as i32),
+            y: ((self.rs_x as u16 as i32) << 16) | (self.rs_y as u16 as i32),
+            flags: ((self.seq as u32) << 24)
+                | ((self.left_trigger as u32) << 16)
+                | ((self.right_trigger as u32) << 8)
+                | (self.pad as u32),
+        }
+    }
+
+    /// Unpack from a [`InputKind::GamepadState`] event; `None` for any other kind.
+    pub fn from_event(ev: &InputEvent) -> Option<GamepadSnapshot> {
+        if ev.kind != InputKind::GamepadState {
+            return None;
+        }
+        Some(GamepadSnapshot {
+            pad: ev.flags as u8,
+            seq: (ev.flags >> 24) as u8,
+            buttons: ev.code,
+            left_trigger: (ev.flags >> 16) as u8,
+            right_trigger: (ev.flags >> 8) as u8,
+            ls_x: (ev.x >> 16) as i16,
+            ls_y: ev.x as i16,
+            rs_x: (ev.y >> 16) as i16,
+            rs_y: ev.y as i16,
+        })
+    }
+
+    /// Fold one per-transition [`GamepadButton`](InputKind::GamepadButton) /
+    /// [`GamepadAxis`](InputKind::GamepadAxis) event into this snapshot (`seq`/`pad` untouched).
+    /// `false` = not a foldable event / unknown axis id (snapshot unchanged).
+    pub fn fold(&mut self, ev: &InputEvent) -> bool {
+        match ev.kind {
+            InputKind::GamepadButton => {
+                if ev.x != 0 {
+                    self.buttons |= ev.code;
+                } else {
+                    self.buttons &= !ev.code;
+                }
+                true
+            }
+            InputKind::GamepadAxis => {
+                let stick = ev.x.clamp(i16::MIN as i32, i16::MAX as i32) as i16;
+                let trigger = ev.x.clamp(0, 255) as u8;
+                match ev.code {
+                    gamepad::AXIS_LS_X => self.ls_x = stick,
+                    gamepad::AXIS_LS_Y => self.ls_y = stick,
+                    gamepad::AXIS_RS_X => self.rs_x = stick,
+                    gamepad::AXIS_RS_Y => self.rs_y = stick,
+                    gamepad::AXIS_LT => self.left_trigger = trigger,
+                    gamepad::AXIS_RT => self.right_trigger = trigger,
+                    _ => return false,
+                }
+                true
+            }
+            _ => false,
+        }
+    }
+
+    /// True when `seq` supersedes `last` (wrapping u8 distance, forward window of 127) — the
+    /// host's reorder gate. `None` (nothing applied yet) always accepts.
+    pub fn seq_newer(seq: u8, last: Option<u8>) -> bool {
+        match last {
+            None => true,
+            Some(l) => (seq.wrapping_sub(l) as i8) > 0,
+        }
+    }
+}
+
 /// A single input event. `#[repr(C)]` — shared verbatim with the C ABI as
 /// `PunktfunkInputEvent`.
 #[repr(C)]
@@ -199,7 +321,79 @@ mod tests {
             };
             assert_eq!(InputEvent::decode(&e.encode()), Some(e));
         }
-        // 12 (one past TouchUp) is not a valid kind.
-        assert_eq!(InputKind::from_u8(12), None);
+        // 13 (one past GamepadState) is not a valid kind.
+        assert_eq!(InputKind::from_u8(13), None);
+    }
+
+    #[test]
+    fn gamepad_snapshot_roundtrip() {
+        let s = GamepadSnapshot {
+            pad: 3,
+            seq: 200,
+            buttons: gamepad::BTN_A | gamepad::BTN_PADDLE4 | gamepad::BTN_MISC1,
+            left_trigger: 255,
+            right_trigger: 1,
+            ls_x: -32768,
+            ls_y: 32767,
+            rs_x: -1,
+            rs_y: 12345,
+        };
+        let ev = s.to_event();
+        assert_eq!(ev.kind, InputKind::GamepadState);
+        // Survives the wire encode/decode unchanged.
+        let dec = InputEvent::decode(&ev.encode()).unwrap();
+        assert_eq!(GamepadSnapshot::from_event(&dec), Some(s));
+        // Non-snapshot kinds unpack to None.
+        let axis = InputEvent {
+            kind: InputKind::GamepadAxis,
+            _pad: [0; 3],
+            code: gamepad::AXIS_LT,
+            x: 255,
+            y: 0,
+            flags: 0,
+        };
+        assert_eq!(GamepadSnapshot::from_event(&axis), None);
+    }
+
+    #[test]
+    fn gamepad_snapshot_fold() {
+        let mut s = GamepadSnapshot::default();
+        let ev = |kind: InputKind, code: u32, x: i32| InputEvent {
+            kind,
+            _pad: [0; 3],
+            code,
+            x,
+            y: 0,
+            flags: 0,
+        };
+        // Button down/up sets and clears its bit.
+        assert!(s.fold(&ev(InputKind::GamepadButton, gamepad::BTN_A, 1)));
+        assert!(s.fold(&ev(InputKind::GamepadButton, gamepad::BTN_RB, 1)));
+        assert_eq!(s.buttons, gamepad::BTN_A | gamepad::BTN_RB);
+        assert!(s.fold(&ev(InputKind::GamepadButton, gamepad::BTN_A, 0)));
+        assert_eq!(s.buttons, gamepad::BTN_RB);
+        // Axes land in their slots; triggers clamp to 0..255, sticks to i16.
+        assert!(s.fold(&ev(InputKind::GamepadAxis, gamepad::AXIS_LT, 300)));
+        assert_eq!(s.left_trigger, 255);
+        assert!(s.fold(&ev(InputKind::GamepadAxis, gamepad::AXIS_LS_Y, -40000)));
+        assert_eq!(s.ls_y, i16::MIN);
+        // Unknown axis / unrelated kind leave the snapshot untouched.
+        assert!(!s.fold(&ev(InputKind::GamepadAxis, 99, 1)));
+        assert!(!s.fold(&ev(InputKind::KeyDown, 30, 1)));
+    }
+
+    #[test]
+    fn gamepad_snapshot_seq_gate() {
+        // First snapshot always applies.
+        assert!(GamepadSnapshot::seq_newer(0, None));
+        // Strictly newer within the forward window applies; equal/older doesn't.
+        assert!(GamepadSnapshot::seq_newer(6, Some(5)));
+        assert!(!GamepadSnapshot::seq_newer(5, Some(5)));
+        assert!(!GamepadSnapshot::seq_newer(4, Some(5)));
+        // Wraps: 2 supersedes 250 (forward distance 8), not the reverse.
+        assert!(GamepadSnapshot::seq_newer(2, Some(250)));
+        assert!(!GamepadSnapshot::seq_newer(250, Some(2)));
+        // Exactly half the window away is treated as stale (i8 > 0 excludes -128).
+        assert!(!GamepadSnapshot::seq_newer(133, Some(5)));
     }
 }

crates/punktfunk-core/src/quic.rs

@@ -137,6 +137,13 @@ pub const QUIT_CLOSE_CODE: u32 = 0x51;
 /// returns to its launcher on session end), so it is purely refinement. Shared so host + clients agree.
 pub const APP_EXITED_CLOSE_CODE: u32 = 0x52;
 
+/// [`Welcome::host_caps`] bit: the host applies [`InputKind::GamepadState`]
+/// (crate::input::InputKind::GamepadState) snapshot events — full per-pad state with a reorder
+/// sequence number. A capable client then sends gamepad state as snapshots (idempotent on the
+/// lossy datagram plane, periodically refreshed) instead of the fragile per-transition
+/// button/axis events; toward a host that doesn't set the bit it keeps the legacy events.
+pub const HOST_CAP_GAMEPAD_STATE: u8 = 0x01;
+
 /// [`Hello::video_codecs`] bit: the client can decode H.264 / AVC. The GPU-less **software**
 /// encode path (openh264) emits H.264, so a client that wants to stream from a software host MUST
 /// advertise this.
@@ -317,6 +324,11 @@ pub struct Welcome {
     /// HEVC). Appended after `audio_channels` as a single trailing byte; an older host that omits it
     /// decodes to [`CODEC_HEVC`] (every pre-negotiation host sent HEVC).
     pub codec: u8,
+    /// Host input capabilities — a bitfield of [`HOST_CAP_GAMEPAD_STATE`]. The client picks the
+    /// wire form its gamepad events take from this (snapshots for a capable host, the legacy
+    /// per-transition events otherwise). Appended after `codec` as a single trailing byte; an
+    /// older host that omits it decodes to `0` (no capabilities — legacy events only).
+    pub host_caps: u8,
 }
 
 /// `client → host`: data plane is bound, begin streaming.
@@ -949,6 +961,8 @@ impl Welcome {
         b.push(self.audio_channels);
         // Resolved video codec at offset 66 — older clients stop before this → HEVC.
         b.push(self.codec);
+        // Host input caps at offset 67 — older clients stop before this → 0 (legacy input only).
+        b.push(self.host_caps);
         b
     }
 
@@ -1031,6 +1045,9 @@ impl Welcome {
                 Some(CODEC_AV1) => CODEC_AV1,
                 _ => CODEC_HEVC,
             },
+            // Optional trailing host-caps byte — absent on an older host → 0 (no gamepad-state
+            // snapshots; the client keeps sending legacy per-transition events).
+            host_caps: b.get(67).copied().unwrap_or(0),
         })
     }
 
@@ -2228,6 +2245,7 @@ mod tests {
             chroma_format: CHROMA_IDC_444,
             audio_channels: 2,
             codec: CODEC_H264, // exercise a non-default codec through the roundtrip
+            host_caps: HOST_CAP_GAMEPAD_STATE,
         };
         assert_eq!(Welcome::decode(&w.encode()).unwrap(), w);
     }
@@ -2329,6 +2347,7 @@ mod tests {
                 chroma_format: CHROMA_IDC_420,
                 audio_channels: 2,
                 codec: CODEC_H264,
+                host_caps: 0,
             }
             .encode(),
         )
@@ -2526,9 +2545,10 @@ mod tests {
             chroma_format: CHROMA_IDC_444,
             audio_channels: 6, // 5.1 — exercises the non-default trailing byte
             codec: CODEC_HEVC,
+            host_caps: HOST_CAP_GAMEPAD_STATE,
         };
         let wenc = w.encode();
-        assert_eq!(wenc.len(), 67); // 60 base + 4 colour + 1 chroma + 1 audio-channels + 1 codec byte
+        assert_eq!(wenc.len(), 68); // 60 base + 4 colour + chroma + audio-channels + codec + host-caps
         let legacy_w = Welcome::decode(&wenc[..53]).unwrap();
         assert_eq!(legacy_w.compositor, CompositorPref::Auto);
         assert_eq!(legacy_w.gamepad, GamepadPref::Auto);
@@ -2571,6 +2591,12 @@ mod tests {
             CHROMA_IDC_444
         ); // full form carries 4:4:4
         assert_eq!(Welcome::decode(&wenc).unwrap().audio_channels, 6); // ...and 5.1
+                                                                       // A pre-host-caps (67-byte) Welcome → 0 (legacy input only); the full form carries the bit.
+        assert_eq!(Welcome::decode(&wenc[..67]).unwrap().host_caps, 0);
+        assert_eq!(
+            Welcome::decode(&wenc).unwrap().host_caps,
+            HOST_CAP_GAMEPAD_STATE
+        );
     }
 
     #[test]

crates/punktfunk-core/src/session.rs

@@ -28,15 +28,18 @@ pub struct Frame {
 /// One end of a stream. Constructed for a single [`Role`]; calling the other role's
 /// methods returns [`PunktfunkError::InvalidArg`].
 ///
-/// Note: the AEAD layer authenticates each datagram but does **not** provide anti-replay.
-/// Video replays are largely absorbed by the reassembler's per-frame dedup, but replayed
-/// input events are not yet filtered. A sliding-window replay filter keyed on the
-/// authenticated sequence belongs with the pairing/handshake layer (the GameStream host); until then,
-/// rely on the LAN/VPN transport assumption (plan §1).
+/// Anti-replay: the receive path runs each opened datagram's AEAD-authenticated sequence through a
+/// sliding-window filter ([`ReplayWindow`]), so a captured, validly-sealed datagram can't be replayed
+/// by an on-path attacker — closing the input-replay gap that previously rested solely on the
+/// LAN/VPN transport assumption (plan §1). Genuine reordering within the window is still accepted;
+/// video additionally benefits from the reassembler's per-frame dedup.
 pub struct Session {
     config: Config,
     coder: Box<dyn ErasureCoder>,
     crypto: Option<SessionCrypto>,
+    /// Anti-replay window over the peer's authenticated sequence (receive side). `Some` exactly when
+    /// `crypto` is — the plaintext probe path carries no sequence to filter on.
+    replay: Option<ReplayWindow>,
     transport: Box<dyn Transport>,
     packetizer: Packetizer,
     reassembler: Reassembler,
@@ -68,11 +71,15 @@ impl Session {
         let crypto = config
             .encrypt
             .then(|| SessionCrypto::new(&config.key, config.salt, config.role));
+        // A receive-side replay window exists exactly when the datagrams are sealed (they carry the
+        // authenticated sequence the window keys on). Both roles receive from their peer.
+        let replay = config.encrypt.then(ReplayWindow::new);
         let packetizer = Packetizer::new(&config);
         let reassembler = Reassembler::new(ReassemblerLimits::from_config(&config));
         Ok(Session {
             coder,
             crypto,
+            replay,
             transport,
             packetizer,
             reassembler,
@@ -130,6 +137,16 @@ impl Session {
         }
     }
 
+    /// Feed an opened datagram's authenticated sequence to the anti-replay window: `true` = fresh
+    /// (accept), `false` = a replay or older than the window (drop). Returns `true` when the session
+    /// isn't encrypting (no window, and no sequence on the wire to key on).
+    fn accept_seq(&mut self, seq: u64) -> bool {
+        match self.replay.as_mut() {
+            Some(w) => w.accept(seq),
+            None => true,
+        }
+    }
+
     // -- Host path --------------------------------------------------------
 
     /// Host: FEC-protect, packetize, and seal one encoded access unit into wire packets WITHOUT
@@ -225,6 +242,13 @@ impl Session {
                 Ok(p) => p,
                 Err(_) => continue, // drop undecryptable noise
             };
+            // Anti-replay: a captured input datagram replayed by an on-path attacker opens cleanly
+            // (its sequence + tag are still valid) — the window is what rejects the second copy.
+            // `len >= 8` is guaranteed because the sealed-path open above succeeded.
+            if self.replay.is_some() && !self.accept_seq(seq_of(&wire)) {
+                StatsCounters::add(&self.stats.packets_dropped, 1);
+                continue;
+            }
             StatsCounters::add(&self.stats.packets_received, 1);
             if let Some(ev) = InputEvent::decode(&pkt) {
                 return Ok(Some(ev));
@@ -276,6 +300,13 @@ impl Session {
                 Ok(p) => p,
                 Err(_) => continue,
             };
+            // Anti-replay (same rationale as poll_input): reject a datagram whose authenticated
+            // sequence was already seen. Video also dedups per-frame downstream, but filtering here
+            // is uniform and cheap. `len >= 8` because the sealed-path open above succeeded.
+            if self.replay.is_some() && !self.accept_seq(seq_of(&self.recv_scratch[i][..len])) {
+                StatsCounters::add(&self.stats.packets_dropped, 1);
+                continue;
+            }
             StatsCounters::add(&self.stats.packets_received, 1);
             StatsCounters::add(&self.stats.bytes_received, pkt.len() as u64);
             // The reassembler validates the packet via its parsed header (`magic`),
@@ -347,3 +378,168 @@ impl Session {
         Ok(())
     }
 }
+
+/// Extract the AEAD-authenticated 8-byte big-endian sequence prefix from a sealed wire datagram.
+/// Only called on the encrypted receive path, where a preceding successful open has already
+/// established `wire.len() >= 8`.
+fn seq_of(wire: &[u8]) -> u64 {
+    u64::from_be_bytes(wire[..8].try_into().unwrap())
+}
+
+/// Depth of the anti-replay window, in sequences. The sender advances its sequence once per
+/// datagram, so at the data plane's packet rate 4096 is roughly 33 ms of reorder tolerance for the
+/// video stream (well beyond any reordering still useful for a live frame) and effectively unbounded
+/// for the sparse input stream — while bounding how far back a replay could hide.
+const REPLAY_WINDOW: u64 = 4096;
+const REPLAY_WORDS: usize = (REPLAY_WINDOW / 64) as usize;
+
+/// Sliding-window anti-replay filter over the AEAD-authenticated wire sequence. The sender counts
+/// its datagrams from 0, and the protocol never legitimately re-sends a sequence (FEC recovery
+/// shards get fresh ones), so a sequence seen twice is a replay. The AEAD tag already authenticates
+/// the sequence — a forged one can't open — so this only has to reject *duplicates* of validly
+/// sealed datagrams (and anything older than the window, which we can no longer prove is fresh).
+/// Genuine reordering within the window is accepted. Bitmap-per-sequence, indexed `seq % WINDOW`.
+struct ReplayWindow {
+    /// Highest sequence accepted so far; `seen` stays false until the first datagram.
+    highest: u64,
+    seen: bool,
+    /// One bit per in-window sequence in `(highest - WINDOW, highest]`.
+    bits: [u64; REPLAY_WORDS],
+}
+
+impl ReplayWindow {
+    fn new() -> ReplayWindow {
+        ReplayWindow {
+            highest: 0,
+            seen: false,
+            bits: [0; REPLAY_WORDS],
+        }
+    }
+
+    #[inline]
+    fn word_bit(seq: u64) -> (usize, u64) {
+        let idx = (seq % REPLAY_WINDOW) as usize;
+        (idx / 64, 1u64 << (idx % 64))
+    }
+    fn is_set(&self, seq: u64) -> bool {
+        let (w, b) = Self::word_bit(seq);
+        self.bits[w] & b != 0
+    }
+    fn set(&mut self, seq: u64) {
+        let (w, b) = Self::word_bit(seq);
+        self.bits[w] |= b;
+    }
+    fn unset(&mut self, seq: u64) {
+        let (w, b) = Self::word_bit(seq);
+        self.bits[w] &= !b;
+    }
+
+    /// Record `seq`, returning `true` if it's fresh (accept) or `false` if it's a replay / too old.
+    fn accept(&mut self, seq: u64) -> bool {
+        if !self.seen {
+            self.seen = true;
+            self.highest = seq;
+            self.set(seq);
+            return true;
+        }
+        if seq > self.highest {
+            // Advance the window. Sequences between the old and new high slide in unseen, so clear
+            // their (possibly stale, from a full window ago) slots — unless we jumped an entire
+            // window, in which case wipe the bitmap wholesale.
+            if seq - self.highest >= REPLAY_WINDOW {
+                self.bits = [0; REPLAY_WORDS];
+            } else {
+                let mut s = self.highest + 1;
+                while s < seq {
+                    self.unset(s);
+                    s += 1;
+                }
+            }
+            self.highest = seq;
+            self.set(seq);
+            true
+        } else if self.highest - seq >= REPLAY_WINDOW || self.is_set(seq) {
+            // Older than the window (can't prove it isn't a replay) or already seen (a duplicate) —
+            // either way, drop it.
+            false
+        } else {
+            self.set(seq); // in-window and not yet seen — a genuine reorder
+            true
+        }
+    }
+}
+
+#[cfg(test)]
+mod replay_tests {
+    use super::*;
+
+    #[test]
+    fn accepts_in_order_and_rejects_duplicates() {
+        let mut w = ReplayWindow::new();
+        for seq in 0..1000 {
+            assert!(w.accept(seq), "fresh in-order seq {seq} must be accepted");
+        }
+        // Every one of those is now a replay.
+        for seq in 0..1000 {
+            assert!(!w.accept(seq), "replayed seq {seq} must be rejected");
+        }
+    }
+
+    #[test]
+    fn accepts_reorder_within_window_once() {
+        let mut w = ReplayWindow::new();
+        assert!(w.accept(100));
+        // Earlier-but-in-window sequences (a genuine reorder) are accepted exactly once.
+        assert!(w.accept(80));
+        assert!(!w.accept(80), "second copy of a reordered seq is a replay");
+        assert!(w.accept(99));
+        assert!(
+            !w.accept(100),
+            "the high-water seq itself can't be replayed"
+        );
+    }
+
+    #[test]
+    fn rejects_older_than_window() {
+        let mut w = ReplayWindow::new();
+        assert!(w.accept(REPLAY_WINDOW * 2));
+        // Anything a full window or more behind the high-water mark is dropped (can't prove fresh).
+        assert!(!w.accept(REPLAY_WINDOW * 2 - REPLAY_WINDOW));
+        assert!(!w.accept(0));
+        // But just inside the window is still accepted.
+        assert!(w.accept(REPLAY_WINDOW * 2 - (REPLAY_WINDOW - 1)));
+    }
+
+    #[test]
+    fn large_forward_jump_wipes_stale_bits() {
+        let mut w = ReplayWindow::new();
+        assert!(w.accept(5));
+        // Jump far forward (more than a window). The slot for an old seq that aliases 5 mod WINDOW
+        // must read as unseen afterward, i.e. the jump cleared it — so a NEW seq there is accepted.
+        let far = 10 * REPLAY_WINDOW + 5;
+        assert!(w.accept(far));
+        assert!(
+            !w.accept(5),
+            "the pre-jump seq is now far older than the window"
+        );
+        // A fresh seq aliasing 5 (mod WINDOW) but inside the new window is accepted, proving the
+        // stale bit was cleared rather than mistaken for a replay.
+        assert!(w.accept(far - REPLAY_WINDOW + 1));
+    }
+
+    #[test]
+    fn first_seq_need_not_be_zero() {
+        // Startup loss can mean the first datagram we ever open isn't seq 0.
+        let mut w = ReplayWindow::new();
+        assert!(w.accept(42));
+        assert!(!w.accept(42));
+        assert!(w.accept(43));
+    }
+
+    #[test]
+    fn seq_of_reads_the_big_endian_prefix() {
+        let mut wire = 0x0102_0304_0506_0708u64.to_be_bytes().to_vec();
+        wire.extend_from_slice(b"ciphertext-and-tag");
+        assert_eq!(seq_of(&wire), 0x0102_0304_0506_0708);
+    }
+}

crates/punktfunk-host/README.md

@@ -73,7 +73,7 @@ src/
   vdisplay/          per-compositor virtual outputs (kwin · gamescope · mutter · wlroots)
   capture/ · capture.rs    screen/dmabuf capture (+ Windows IDD-push)
   encode/ · encode.rs      per-GPU encoders (nvenc · vaapi · ffmpeg_win (AMF/QSV) · sw)
-  zerocopy/          dmabuf → CUDA → NVENC bridges (EGL/GL tiled, Vulkan LINEAR)
+  linux/zerocopy/    dmabuf → CUDA → NVENC bridges (EGL/GL tiled, Vulkan LINEAR)
   inject/ · inject.rs      input backends (libei · wlr · uinput gamepads · UHID DualSense/DS4)
   audio/ · audio.rs        Opus out + virtual mic (PipeWire / WASAPI)
   gamestream/        Moonlight compat: nvhttp · pairing · rtsp · control · stream · gamepad · apps
@@ -88,4 +88,4 @@ src/
 - **[`punktfunk-core`](../punktfunk-core/README.md)** — the shared protocol · FEC · crypto core
 - **[Clients](../../clients/)** — the apps that connect (Apple · Linux · Windows · Android · probe)
 - **[Packaging](../../packaging/README.md)** & **[docs](https://docs.punktfunk.unom.io)** — install & operate
-- **[`design/`](../../design/README.md)** — architecture rationale and deep-dive plans
+- **punktfunk-planning** (internal planning repo) — architecture rationale and deep-dive plans

crates/punktfunk-host/src/capture/windows/idd_push.rs

@@ -760,6 +760,56 @@ impl IddPushCapturer {
         target: WinCaptureTarget,
         preferred: Option<(u32, u32, u32)>,
         client_10bit: bool,
+    ) -> Result<Self> {
+        // The ring MUST live on the adapter the driver's swap-chain renders on. Primary: the
+        // selected render GPU — the same pick SET_RENDER_ADAPTER pinned the driver to at monitor
+        // ADD, so on a healthy box they agree, and NVENC gets a device on a real GPU adapter.
+        // (`target.adapter_luid` is NOT that adapter: the ADD reply carries
+        // `IDARG_OUT_MONITORARRIVAL.OsAdapterLuid` = the IddCx DISPLAY adapter — verified
+        // on-glass; it stays a last-resort fallback for a pickerless box only.) When the pick and
+        // the driver HAVE drifted — identical twin GPUs whose max-VRAM tie moved between ADD and
+        // this open, or a stale kept monitor across an adapter re-init — the driver reports
+        // TEX_FAIL plus the adapter it actually renders on, and the rebind below reopens on that.
+        let luid = crate::win_adapter::resolve_render_adapter_luid().unwrap_or(LUID {
+            LowPart: (target.adapter_luid & 0xffff_ffff) as u32,
+            HighPart: (target.adapter_luid >> 32) as i32,
+        });
+        match Self::open_on(target.clone(), preferred, client_10bit, luid) {
+            Ok(me) => Ok(me),
+            Err(e) => {
+                // Self-heal a render-adapter mismatch ONCE: on TEX_FAIL the driver has reported the
+                // adapter its swap-chain ACTUALLY renders on (a stale monitor across an adapter
+                // re-init, or a driver that ignored SET_RENDER_ADAPTER). Rebinding the ring to that
+                // adapter beats failing the session — the outer pipeline retries would repeat the
+                // exact same mismatch.
+                let driver_luid = e
+                    .downcast_ref::<AttachTexFail>()
+                    .map(|tf| tf.driver_luid)
+                    .filter(|d| *d != 0 && *d != crate::capture::dxgi::pack_luid(luid));
+                let Some(packed) = driver_luid else {
+                    return Err(e);
+                };
+                let drv = LUID {
+                    LowPart: (packed & 0xffff_ffff) as u32,
+                    HighPart: (packed >> 32) as i32,
+                };
+                tracing::warn!(
+                    ring_adapter = format!("{:08x}:{:08x}", luid.HighPart, luid.LowPart),
+                    driver_adapter = format!("{:08x}:{:08x}", drv.HighPart, drv.LowPart),
+                    "IDD push: ring/driver render-adapter mismatch — rebinding the ring to the \
+                     driver's reported adapter"
+                );
+                Self::open_on(target, preferred, client_10bit, drv)
+                    .context("IDD-push rebind to the driver's reported render adapter")
+            }
+        }
+    }
+
+    fn open_on(
+        target: WinCaptureTarget,
+        preferred: Option<(u32, u32, u32)>,
+        client_10bit: bool,
+        luid: LUID,
     ) -> Result<Self> {
         let (pw, ph, _hz) = preferred
             .context("IDD push needs the negotiated mode (WxH) to size the shared ring")?;
@@ -829,10 +879,10 @@ impl IddPushCapturer {
             } else {
                 DXGI_FORMAT_B8G8R8A8_UNORM
             };
-            // Create our device on the discrete render GPU (where NVENC runs); the driver must render
-            // the swap-chain on the SAME adapter for the shared textures to open (it reports its actual
-            // render LUID into the header so we can detect a mismatch).
-            let luid = resolve_render_adapter_luid_or(target.adapter_luid);
+            // Our device (ring + zero-copy NVENC) lives on `luid` — the selected render GPU per
+            // `open_inner`; the driver must render the swap-chain on the SAME adapter for the
+            // shared textures to open (it reports its actual render LUID into the header, which
+            // `open_inner` uses to rebind once if this mismatches).
             let factory: IDXGIFactory4 = CreateDXGIFactory1().context("CreateDXGIFactory1")?;
             let adapter: IDXGIAdapter1 = factory
                 .EnumAdapterByLuid(luid)
@@ -991,13 +1041,31 @@ impl IddPushCapturer {
             // diagnostics — the real handshake is the atomic `magic`/`latest` (same access as
             // log_driver_status_once).
             let st = unsafe { (*self.header).driver_status };
-            if matches!(st, DRV_STATUS_TEX_FAIL | DRV_STATUS_NO_DEVICE1) {
+            if st == DRV_STATUS_TEX_FAIL {
+                // SAFETY: as above — in-bounds, aligned word reads of best-effort diagnostic fields
+                // through the owned, live header mapping; no reference into the shared region is formed.
+                // The driver wrote its render LUID BEFORE attempting the texture opens
+                // (frame_transport.rs step 2), so it is valid here.
+                let (detail, lo, hi) = unsafe {
+                    (
+                        (*self.header).driver_status_detail,
+                        (*self.header).driver_render_luid_low,
+                        (*self.header).driver_render_luid_high,
+                    )
+                };
+                // Typed so `open_inner` can rebind the ring to the driver's adapter once.
+                return Err(anyhow::Error::new(AttachTexFail {
+                    detail,
+                    driver_luid: ((hi as i64) << 32) | (lo as i64 & 0xffff_ffff),
+                }));
+            }
+            if st == DRV_STATUS_NO_DEVICE1 {
                 // SAFETY: as above — an in-bounds, aligned `u32` read of a best-effort diagnostic field
                 // through the owned, live header mapping; no reference into the shared region is formed.
                 let detail = unsafe { (*self.header).driver_status_detail };
                 bail!(
                     "IDD-push driver failed to attach (driver_status={st} detail=0x{detail:08x} — \
-                     render-adapter mismatch?)"
+                     the driver has no ID3D11Device1 to open shared resources)"
                 );
             }
             // Attached AND a frame has been published — the publish token's seq advances past 0.
@@ -1415,17 +1483,31 @@ impl IddPushCapturer {
     }
 }
 
-/// The selected render GPU LUID (where the encoder runs), falling back to the monitor's `OsAdapterLuid`.
-fn resolve_render_adapter_luid_or(fallback_packed: i64) -> LUID {
-    if let Some(l) = crate::win_adapter::resolve_render_adapter_luid() {
-        return l;
-    }
-    LUID {
-        LowPart: (fallback_packed & 0xffff_ffff) as u32,
-        HighPart: (fallback_packed >> 32) as i32,
+/// `wait_for_attach`'s DRV_STATUS_TEX_FAIL as a typed error: the driver could not open the ring
+/// textures, and `driver_luid` (packed, from the shared header) is the adapter its swap-chain
+/// ACTUALLY renders on — `open_inner` downcasts to this to rebind the ring there once.
+#[derive(Debug)]
+struct AttachTexFail {
+    detail: u32,
+    driver_luid: i64,
+}
+
+impl std::fmt::Display for AttachTexFail {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(
+            f,
+            "IDD-push driver failed to attach (driver_status={DRV_STATUS_TEX_FAIL} \
+             detail=0x{:08x}): it could not open the ring textures — its swap-chain renders on \
+             adapter {:08x}:{:08x}, not the ring's (render-adapter mismatch)",
+            self.detail,
+            (self.driver_luid >> 32) as i32,
+            (self.driver_luid & 0xffff_ffff) as u32,
+        )
     }
 }
 
+impl std::error::Error for AttachTexFail {}
+
 impl Capturer for IddPushCapturer {
     fn next_frame(&mut self) -> Result<CapturedFrame> {
         let deadline = Instant::now() + Duration::from_secs(20);

crates/punktfunk-host/src/config.rs

@@ -68,6 +68,13 @@ pub struct HostConfig {
     /// backend (the legacy SudoVDA backend was removed), so this is currently informational — kept for the
     /// shipped `host.env` and as a forward seam if a second backend is ever added.
     pub vdisplay: Option<String>,
+    /// `PUNKTFUNK_RECOVER_SESSION_CMD` — operator hook fired (debounced) when a client connects while NO
+    /// graphical session is live for this uid: the state a compositor crash leaves behind (gnome-shell
+    /// SIGSEGV → GDM greeter, whose auto-login is once-per-boot, so the box would otherwise need a walk-up
+    /// or reboot). Typically `sudo -n systemctl restart gdm` with a matching NOPASSWD sudoers rule, or
+    /// `systemctl restart display-manager` under a polkit rule — with auto-login enabled the restart brings
+    /// the desktop back and the client's retry lands in it. Unset/empty = disabled (the default).
+    pub recover_session_cmd: Option<String>,
 }
 
 impl HostConfig {
@@ -101,6 +108,8 @@ impl HostConfig {
             compositor: val("PUNKTFUNK_COMPOSITOR"),
             gamepad: val("PUNKTFUNK_GAMEPAD"),
             vdisplay: val("PUNKTFUNK_VDISPLAY"),
+            recover_session_cmd: val("PUNKTFUNK_RECOVER_SESSION_CMD")
+                .filter(|s| !s.trim().is_empty()),
         }
     }
 }

crates/punktfunk-host/src/gpu.rs

@@ -91,7 +91,8 @@ impl GpuInfo {
 }
 
 /// Assign the stable `id` + `occurrence` fields after enumeration (occurrence = index among
-/// same-(vendor,device) twins, in enumeration order).
+/// same-(vendor,device) twins, in inventory order — Windows sorts the inventory by LUID first so
+/// twin numbering is stable for the boot, see [`enumerate`]).
 fn assign_ids(gpus: &mut [GpuInfo]) {
     for i in 0..gpus.len() {
         let occ = gpus[..i]
@@ -268,6 +269,13 @@ pub(crate) fn enumerate() -> Vec<GpuInfo> {
             });
         }
     }
+    // Deterministic inventory order. DXGI enumerates the adapter owning the primary display first,
+    // and the vdisplay flow itself MOVES the primary mid-session (exclusive mode turns the physical
+    // screens off) — which flipped every order-relative selection between IDENTICAL twin GPUs from
+    // one call to the next (the max-VRAM tie, the env-substring first match, occurrence numbering):
+    // monitor ADD pinned the driver to one twin, the ring open picked the other → TEX_FAIL. LUIDs
+    // are creation-ordered and stable for the boot, so sorting by them makes selection stable too.
+    out.sort_by_key(|g| ((g.handle.luid_high as i64) << 32) | g.handle.luid_low as i64);
     assign_ids(&mut out);
     out
 }

crates/punktfunk-host/src/inject/linux/kwin_fake_input.rs

@@ -425,7 +425,7 @@ impl InputInjector for KwinFakeInjector {
                 self.fake.touch_frame();
             }
             // Gamepads are injected through uinput, not the compositor.
-            InputKind::GamepadButton | InputKind::GamepadAxis => {}
+            InputKind::GamepadState | InputKind::GamepadButton | InputKind::GamepadAxis => {}
         }
         // Surface protocol errors / disconnects, then push the batch to the compositor.
         self.queue

crates/punktfunk-host/src/inject/linux/libei.rs

@@ -403,6 +403,7 @@ fn kind_bit(kind: InputKind) -> u32 {
         InputKind::TouchUp => 9,
         InputKind::GamepadButton => 10,
         InputKind::GamepadAxis => 11,
+        InputKind::GamepadState => 12,
     };
     1 << i
 }
@@ -545,7 +546,7 @@ impl EiState {
             InputKind::TouchDown | InputKind::TouchMove | InputKind::TouchUp => {
                 DeviceCapability::Touch
             }
-            InputKind::GamepadButton | InputKind::GamepadAxis => return, // uinput path (later)
+            InputKind::GamepadState | InputKind::GamepadButton | InputKind::GamepadAxis => return, // uinput path (later)
         };
         self.injected += 1;
         let n = self.injected;
@@ -692,7 +693,9 @@ impl EiState {
                 Some(t) => t.up(ev.code),
                 None => emitted = false,
             },
-            InputKind::GamepadButton | InputKind::GamepadAxis => emitted = false,
+            InputKind::GamepadState | InputKind::GamepadButton | InputKind::GamepadAxis => {
+                emitted = false
+            }
         }
 
         if emitted {

crates/punktfunk-host/src/inject/linux/wlr.rs

@@ -254,7 +254,7 @@ impl InputInjector for WlrootsInjector {
                     tracing::debug!(vk = event.code, "unmapped VK keycode — dropped");
                 }
             }
-            InputKind::GamepadButton | InputKind::GamepadAxis => {} // not yet injected
+            InputKind::GamepadState | InputKind::GamepadButton | InputKind::GamepadAxis => {} // not yet injected
             // wlroots has no virtual-touch protocol wired here; touch is the libei path only.
             InputKind::TouchDown | InputKind::TouchMove | InputKind::TouchUp => {}
         }

crates/punktfunk-host/src/inject/windows/sendinput.rs

@@ -297,9 +297,10 @@ impl InputInjector for SendInputInjector {
                 };
                 self.send(&[key(ki)])
             }
-            // Gamepad goes through ViGEm (separate backend). Touch: no SendInput equivalent -> no-op.
+            // Gamepad goes through the XUSB backend. Touch: no SendInput equivalent -> no-op.
             InputKind::GamepadButton
             | InputKind::GamepadAxis
+            | InputKind::GamepadState
             | InputKind::TouchDown
             | InputKind::TouchMove
             | InputKind::TouchUp => Ok(()),

crates/punktfunk-host/src/native_pairing.rs

@@ -84,12 +84,24 @@ struct Pending {
     /// A live parked knock is a genuine device waiting for the operator — eviction skips it unless
     /// every entry is parked, so a cert-rotating flood can't evict the device being onboarded (#13).
     parked: bool,
+    /// Generation of the MOST RECENT knock for this fingerprint. A re-knock bumps it (and wakes
+    /// waiters), so a stale parked connection resolves [`PairingDecision::Superseded`] instead of
+    /// being admitted alongside the newest one — one Approve must admit exactly ONE session.
+    /// (Observed live: a client retried 3× while parked, one console Approve admitted all three,
+    /// and the three concurrent Mutter virtual monitors segfaulted gnome-shell.)
+    knock_seq: u32,
 }
 
 #[derive(Default)]
 struct PendingState {
     next_id: u32,
     items: Vec<Pending>,
+    /// Fingerprint → the knock generation an approval admitted, kept briefly after [`NativePairing::add`]
+    /// clears the pending entry. Closes the last double-admit window: a superseded waiter that only
+    /// polls AFTER the approval (entry gone, fingerprint paired) can't tell it lost from the entry
+    /// alone — this marker lets it resolve `Superseded` instead of a second `Approved`. Pruned on
+    /// the pending TTL and overwritten per fingerprint, so it stays a handful of tuples.
+    admitted: Vec<(String, u32, Instant)>,
 }
 
 /// A pending-approval snapshot for the management API / web console.
@@ -114,6 +126,10 @@ pub enum PairingDecision {
     Denied,
     /// No decision within the wait window — reject; the device can knock again.
     TimedOut,
+    /// A NEWER knock from the same fingerprint replaced this one — close this connection; the
+    /// newest parked connection is the one an approval admits (a retrying client abandons its
+    /// older attempts, and admitting them all crashes compositors — see [`Pending::knock_seq`]).
+    Superseded,
 }
 
 /// Pending knocks older than this are dropped (the device retries; a stale entry shouldn't be
@@ -353,9 +369,25 @@ impl NativePairing {
                 return Err(e);
             }
         }
-        // A device that knocked and is now paired shouldn't linger in the approval list.
+        // A device that knocked and is now paired shouldn't linger in the approval list. Record
+        // WHICH knock generation this pairing admits before clearing the entry: only the waiter
+        // holding that generation may return `Approved`; a superseded sibling that polls after the
+        // clear resolves `Superseded` off this marker (exactly-one-admission — see `admitted`).
         {
             let mut pending = self.pending.lock().unwrap();
+            let admitted_seq = pending
+                .items
+                .iter()
+                .find(|p| p.fp_hex.eq_ignore_ascii_case(fp_hex))
+                .map(|p| p.knock_seq);
+            if let Some(seq) = admitted_seq {
+                pending
+                    .admitted
+                    .retain(|(fp, _, _)| !fp.eq_ignore_ascii_case(fp_hex));
+                pending
+                    .admitted
+                    .push((fp_hex.to_string(), seq, Instant::now()));
+            }
             pending
                 .items
                 .retain(|p| !p.fp_hex.eq_ignore_ascii_case(fp_hex));
@@ -394,11 +426,16 @@ impl NativePairing {
 
     // -- Delegated approval (roadmap §8b-1) --------------------------------
 
-    /// Drop expired pending knocks (called under the lock, mirroring [`Self::expire`]).
+    /// Drop expired pending knocks (called under the lock, mirroring [`Self::expire`]). The
+    /// admitted-generation markers share the TTL — they only matter while a superseded waiter
+    /// could still be parked, which is bounded by the approval wait (well under the TTL).
     fn expire_pending(pending: &mut PendingState) {
         pending
             .items
             .retain(|p| p.requested_at.elapsed() < PENDING_TTL);
+        pending
+            .admitted
+            .retain(|(_, _, at)| at.elapsed() < PENDING_TTL);
     }
 
     /// Pick the entry to evict, optionally restricted to a single source IP: the least-recently-active
@@ -419,12 +456,14 @@ impl NativePairing {
     }
 
     /// Record an unpaired device's knock for delegated approval. Re-knocks from the same fingerprint
-    /// refresh the existing entry in place (same id; a connect-retry loop must not spam the list). A
+    /// refresh the existing entry in place (same id; a connect-retry loop must not spam the list) and
+    /// bump its knock generation — the returned generation is what [`Self::wait_for_decision`] admits,
+    /// so the NEWEST connection wins and any older parked sibling resolves `Superseded`. A
     /// fresh fingerprint gets a new id; the queue is bounded two ways so a flood can't crowd out a
     /// genuine knock (#13): a **per-source-IP cap** ([`MAX_PENDING_PER_IP`]) means one host can hold at
     /// most a few slots, and the global [`PENDING_CAP`] evicts the least-recently-active **non-parked**
     /// entry (never a live, held-open parked knock). The name is sanitized (untrusted).
-    pub fn note_pending(&self, name: &str, fp_hex: &str, src_ip: Option<IpAddr>) {
+    pub fn note_pending(&self, name: &str, fp_hex: &str, src_ip: Option<IpAddr>) -> u32 {
         let name = sanitize_device_name(name, fp_hex);
         let mut pending = self.pending.lock().unwrap();
         Self::expire_pending(&mut pending);
@@ -438,8 +477,19 @@ impl NativePairing {
             if p.src_ip.is_none() {
                 p.src_ip = src_ip;
             }
-            return;
+            p.knock_seq = p.knock_seq.wrapping_add(1);
+            let seq = p.knock_seq;
+            drop(pending);
+            // Wake the previous knock's parked waiter so it sees it was superseded NOW instead of
+            // holding its dead connection open until the approval window lapses.
+            self.changed.notify_waiters();
+            return seq;
         }
+        // A fresh knock lifecycle: drop any admitted-generation marker left from a previous
+        // pair→unpair round of this fingerprint, or it would wrongly supersede the new waiter.
+        pending
+            .admitted
+            .retain(|(fp, _, _)| !fp.eq_ignore_ascii_case(fp_hex));
         // Per-source-IP cap: a single host can't occupy more than MAX_PENDING_PER_IP slots — evict its
         // own oldest entry first so it can't crowd out other devices' knocks (#13).
         if let Some(ip) = src_ip {
@@ -471,22 +521,47 @@ impl NativePairing {
             requested_at: Instant::now(),
             src_ip,
             parked: false,
+            knock_seq: 0,
         });
+        0
     }
 
     /// Mark/unmark the pending entry for `fp_hex` as having a live parked waiter (no-op if it's gone).
-    /// A parked entry is protected from eviction under load (#13).
-    fn set_parked(&self, fp_hex: &str, parked: bool) {
+    /// A parked entry is protected from eviction under load (#13). Gated on `knock_seq` so a
+    /// superseded waiter's exit can't unmark the flag the NEWER waiter (a bumped generation) owns.
+    fn set_parked(&self, fp_hex: &str, knock_seq: u32, parked: bool) {
         let mut pending = self.pending.lock().unwrap();
         if let Some(p) = pending
             .items
             .iter_mut()
-            .find(|p| p.fp_hex.eq_ignore_ascii_case(fp_hex))
+            .find(|p| p.fp_hex.eq_ignore_ascii_case(fp_hex) && p.knock_seq == knock_seq)
         {
             p.parked = parked;
         }
     }
 
+    /// The current knock generation for `fp_hex`, `None` when no entry is pending. A parked waiter
+    /// compares this against its own generation to detect it was superseded by a re-knock.
+    fn knock_seq_of(&self, fp_hex: &str) -> Option<u32> {
+        let pending = self.pending.lock().unwrap();
+        pending
+            .items
+            .iter()
+            .find(|p| p.fp_hex.eq_ignore_ascii_case(fp_hex))
+            .map(|p| p.knock_seq)
+    }
+
+    /// The knock generation the approval of `fp_hex` admitted, if one was recorded (see
+    /// [`PendingState::admitted`]).
+    fn admitted_seq(&self, fp_hex: &str) -> Option<u32> {
+        let pending = self.pending.lock().unwrap();
+        pending
+            .admitted
+            .iter()
+            .find(|(fp, _, _)| fp.eq_ignore_ascii_case(fp_hex))
+            .map(|(_, seq, _)| *seq)
+    }
+
     /// The devices currently awaiting approval (for the management API).
     pub fn pending(&self) -> Vec<PendingRequest> {
         let mut pending = self.pending.lock().unwrap();
@@ -560,29 +635,41 @@ impl NativePairing {
     }
 
     /// Park (async) until an operator decides on a knock identified by `fp_hex`, up to `timeout`.
+    /// `knock_seq` is the generation [`Self::note_pending`] returned for THIS connection's knock.
     /// Returns [`PairingDecision::Approved`] the instant the fingerprint is paired (console
-    /// approve or a concurrent PIN ceremony), [`PairingDecision::Denied`] if its pending entry is
-    /// dropped without pairing, or [`PairingDecision::TimedOut`] if the window lapses. Holds no
-    /// lock across the await. The QUIC accept path calls this right after [`Self::note_pending`]
-    /// to keep the knocking connection open until a human clicks Approve — so the device pairs and
-    /// streams with no reconnect (delegated approval, roadmap §8b-1).
-    pub async fn wait_for_decision(&self, fp_hex: &str, timeout: Duration) -> PairingDecision {
+    /// approve or a concurrent PIN ceremony), [`PairingDecision::Superseded`] the instant a newer
+    /// knock from the same fingerprint replaces this one (a retrying client — only the newest
+    /// connection is admitted; three siblings admitted at once has crashed gnome-shell live),
+    /// [`PairingDecision::Denied`] if its pending entry is dropped without pairing, or
+    /// [`PairingDecision::TimedOut`] if the window lapses. Holds no lock across the await. The
+    /// QUIC accept path calls this right after [`Self::note_pending`] to keep the knocking
+    /// connection open until a human clicks Approve — so the device pairs and streams with no
+    /// reconnect (delegated approval, roadmap §8b-1).
+    pub async fn wait_for_decision(
+        &self,
+        fp_hex: &str,
+        knock_seq: u32,
+        timeout: Duration,
+    ) -> PairingDecision {
         // Mark this knock parked so a cert-rotating flood can't evict the genuine, held-open
         // connection out of the pending queue while the operator decides (#13). Cleared on every
-        // exit path by the guard's Drop.
-        self.set_parked(fp_hex, true);
+        // exit path by the guard's Drop (generation-gated, so a superseded waiter's exit never
+        // unmarks the newer waiter's flag).
+        self.set_parked(fp_hex, knock_seq, true);
         struct ParkGuard<'a> {
             np: &'a NativePairing,
             fp: &'a str,
+            seq: u32,
         }
         impl Drop for ParkGuard<'_> {
             fn drop(&mut self) {
-                self.np.set_parked(self.fp, false);
+                self.np.set_parked(self.fp, self.seq, false);
             }
         }
         let _park = ParkGuard {
             np: self,
             fp: fp_hex,
+            seq: knock_seq,
         };
         let deadline = tokio::time::Instant::now() + timeout;
         loop {
@@ -592,16 +679,33 @@ impl NativePairing {
             tokio::pin!(notified);
             notified.as_mut().enable();
 
+            // Superseded check FIRST: once a newer knock owns the fingerprint, this connection
+            // must never be admitted — not even if the approval lands before we wake.
+            match self.knock_seq_of(fp_hex) {
+                Some(cur) if cur != knock_seq => return PairingDecision::Superseded,
+                _ => {}
+            }
             if self.is_paired(fp_hex) {
-                return PairingDecision::Approved;
+                // Paired with the pending entry already cleared: make sure the approval admitted
+                // OUR generation. A superseded waiter that first polls after `add()` sees the same
+                // paired/no-entry state as the winner — the admitted marker breaks the tie.
+                match self.admitted_seq(fp_hex) {
+                    Some(adm) if adm != knock_seq => return PairingDecision::Superseded,
+                    _ => return PairingDecision::Approved,
+                }
             }
             if !self.pending_contains(fp_hex) {
                 // Neither pending nor paired. This is almost always a denial — but it can also be
                 // the tiny interval inside `add()` between pinning and clearing the pending entry.
                 // Re-check `is_paired` once: because `add()` pins BEFORE it clears pending, a
-                // cleared-pending observation that is really an approval will now read as paired.
+                // cleared-pending observation that is really an approval will now read as paired —
+                // with the same generation tie-break as above (the admitted marker is written in
+                // the same critical section that clears the entry).
                 if self.is_paired(fp_hex) {
-                    return PairingDecision::Approved;
+                    match self.admitted_seq(fp_hex) {
+                        Some(adm) if adm != knock_seq => return PairingDecision::Superseded,
+                        _ => return PairingDecision::Approved,
+                    }
                 }
                 return PairingDecision::Denied;
             }
@@ -781,19 +885,19 @@ mod tests {
         let np = Arc::new(NativePairing::load_with(Some(p.clone()), None, false).unwrap());
 
         // TimedOut: a parked knock with no decision returns TimedOut; the entry survives.
-        np.note_pending("Knocker", "ab01", None);
+        let seq = np.note_pending("Knocker", "ab01", None);
         let d = np
-            .wait_for_decision("ab01", Duration::from_millis(80))
+            .wait_for_decision("ab01", seq, Duration::from_millis(80))
             .await;
         assert_eq!(d, PairingDecision::TimedOut);
         assert!(np.pending_contains("ab01"));
 
         // Approved: approving WHILE parked wakes the waiter with Approved.
         let np2 = np.clone();
-        let waiter =
-            tokio::spawn(
-                async move { np2.wait_for_decision("ab01", Duration::from_secs(5)).await },
-            );
+        let waiter = tokio::spawn(async move {
+            np2.wait_for_decision("ab01", seq, Duration::from_secs(5))
+                .await
+        });
         tokio::time::sleep(Duration::from_millis(30)).await;
         let id = np
             .pending()
@@ -806,12 +910,12 @@ mod tests {
         assert!(np.is_paired("ab01"));
 
         // Denied: denying WHILE parked wakes the waiter with Denied (not held until timeout).
-        np.note_pending("Knock2", "cd02", None);
+        let seq = np.note_pending("Knock2", "cd02", None);
         let np3 = np.clone();
-        let waiter =
-            tokio::spawn(
-                async move { np3.wait_for_decision("cd02", Duration::from_secs(5)).await },
-            );
+        let waiter = tokio::spawn(async move {
+            np3.wait_for_decision("cd02", seq, Duration::from_secs(5))
+                .await
+        });
         tokio::time::sleep(Duration::from_millis(30)).await;
         let id = np
             .pending()
@@ -823,12 +927,67 @@ mod tests {
         assert_eq!(waiter.await.unwrap(), PairingDecision::Denied);
         assert!(!np.is_paired("cd02"));
 
-        // Already paired before the call → immediate Approved (no waiting).
-        let d = np.wait_for_decision("ab01", Duration::from_secs(5)).await;
+        // Already paired before the call (the PIN-ceremony race) → immediate Approved: the ab01
+        // marker admitted generation 0, which is also what a fresh coincidental waiter holds.
+        let d = np
+            .wait_for_decision("ab01", 0, Duration::from_secs(5))
+            .await;
         assert_eq!(d, PairingDecision::Approved);
         let _ = std::fs::remove_file(&p);
     }
 
+    /// One Approve must admit exactly ONE session: a re-knock supersedes the previous parked
+    /// waiter (it resolves `Superseded` immediately, not at timeout), the console list keeps a
+    /// single entry, and a stale-generation waiter that polls only AFTER the approval still
+    /// resolves `Superseded` off the admitted marker. (Live failure this pins down: a client
+    /// knocked 3×, one Approve admitted all three, and the three concurrent Mutter virtual
+    /// monitors segfaulted gnome-shell.)
+    #[tokio::test]
+    async fn newest_knock_supersedes_parked_waiter() {
+        use std::sync::Arc;
+        let p = temp();
+        let _ = std::fs::remove_file(&p);
+        let np = Arc::new(NativePairing::load_with(Some(p.clone()), None, false).unwrap());
+
+        let seq1 = np.note_pending("iPad Pro", "ee01", None);
+        let np1 = np.clone();
+        let waiter1 = tokio::spawn(async move {
+            np1.wait_for_decision("ee01", seq1, Duration::from_secs(5))
+                .await
+        });
+        tokio::time::sleep(Duration::from_millis(30)).await;
+
+        // The device retries: same fingerprint, new connection. The old waiter is superseded at
+        // once; the pending list still shows ONE entry.
+        let seq2 = np.note_pending("iPad Pro", "ee01", None);
+        assert_ne!(seq1, seq2);
+        assert_eq!(waiter1.await.unwrap(), PairingDecision::Superseded);
+        assert_eq!(np.pending().len(), 1);
+
+        let np2 = np.clone();
+        let waiter2 = tokio::spawn(async move {
+            np2.wait_for_decision("ee01", seq2, Duration::from_secs(5))
+                .await
+        });
+        tokio::time::sleep(Duration::from_millis(30)).await;
+        let id = np
+            .pending()
+            .into_iter()
+            .find(|x| x.fingerprint == "ee01")
+            .unwrap()
+            .id;
+        np.approve_pending(id, None).unwrap().unwrap();
+        assert_eq!(waiter2.await.unwrap(), PairingDecision::Approved);
+
+        // A stale-generation waiter polling only after the approval (entry cleared, fingerprint
+        // paired) must NOT read as a second Approved — the admitted marker resolves the tie.
+        let d = np
+            .wait_for_decision("ee01", seq1, Duration::from_millis(80))
+            .await;
+        assert_eq!(d, PairingDecision::Superseded);
+        let _ = std::fs::remove_file(&p);
+    }
+
     /// #9: a window can be bound to one operator-selected fingerprint, so an unrelated (attacker)
     /// fingerprint can neither pair nor BURN the window (it's rejected without a PIN).
     #[test]
@@ -873,8 +1032,8 @@ mod tests {
         // A genuine knock from a different IP, parked (a live held-open connection), survives a flood
         // from many distinct IPs that fills the global cap.
         let legit = IpAddr::from([192, 168, 1, 50]);
-        np.note_pending("Living Room", "legit01", Some(legit));
-        np.set_parked("legit01", true);
+        let seq = np.note_pending("Living Room", "legit01", Some(legit));
+        np.set_parked("legit01", seq, true);
         for i in 0..(PENDING_CAP * 2) {
             let ip = IpAddr::from([10, 0, (i / 256) as u8, (i % 256) as u8]);
             np.note_pending("flood2", &format!("g{i:04}"), Some(ip));

crates/punktfunk-host/src/punktfunk1.rs

@@ -700,13 +700,16 @@ async fn serve_session(
             tracing::info!(name = %label, fingerprint = %fp_hex,
                 "unpaired device knocked — parking connection for delegated approval in the console");
             // Record the QUIC-validated source IP so the pending queue's per-source cap can stop one
-            // host from flooding/evicting genuine knocks (#13).
-            np.note_pending(&label, &fp_hex, Some(peer.ip()));
+            // host from flooding/evicting genuine knocks (#13). The returned knock generation makes
+            // this connection the ONE an approval admits — a retrying client parks a fresh
+            // connection per knock, and admitting every parked sibling on a single Approve spun up
+            // three concurrent Mutter virtual monitors and segfaulted gnome-shell (2026-07-10).
+            let knock_seq = np.note_pending(&label, &fp_hex, Some(peer.ip()));
             // Free the session slot while a human decides — a parked knock must not hold an NVENC
             // permit (a handful of parked knocks would otherwise block every real session).
             drop(permit);
             let decision = tokio::select! {
-                d = np.wait_for_decision(&fp_hex, PENDING_APPROVAL_WAIT) => d,
+                d = np.wait_for_decision(&fp_hex, knock_seq, PENDING_APPROVAL_WAIT) => d,
                 // The client gave up (closed the connection) before a decision — stop waiting.
                 _ = conn.closed() => anyhow::bail!("client disconnected before pairing approval"),
             };
@@ -720,6 +723,10 @@ async fn serve_session(
                     "pairing request not approved within {PENDING_APPROVAL_WAIT:?} \
                      — the device can knock again"
                 ),
+                PairingDecision::Superseded => anyhow::bail!(
+                    "parked knock superseded by a newer connection from the same device — \
+                     only the newest is admitted on approval"
+                ),
             }
             // Re-acquire a session slot for the now-approved session (waits if all slots are busy,
             // exactly like any freshly accepted client).
@@ -1038,6 +1045,9 @@ async fn serve_session(
             // HEVC-precedence tie-break). The client builds its decoder from this instead of
             // assuming HEVC.
             codec: codec_bit,
+            // This host applies sequence-gated gamepad-state snapshots (InputKind::GamepadState),
+            // so capable clients send those instead of the loss-fragile per-transition events.
+            host_caps: punktfunk_core::quic::HOST_CAP_GAMEPAD_STATE,
         };
         io::write_msg(&mut send, &welcome.encode()).await?;
 
@@ -1537,7 +1547,8 @@ async fn serve_session(
 
 /// Per-pad accumulated state: punktfunk/1 gamepad events are incremental (one button or axis
 /// per datagram, see `punktfunk_core::input::gamepad`), the virtual xpad applies full frames.
-#[derive(Clone, Copy, Default)]
+/// A snapshot-capable client replaces the whole state at once ([`PadState::set_snapshot`]).
+#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
 struct PadState {
     buttons: u32,
     left_trigger: u8,
@@ -1574,6 +1585,17 @@ impl PadState {
         true
     }
 
+    /// Replace the whole state from one client snapshot (the [`InputKind::GamepadState`] form).
+    fn set_snapshot(&mut self, s: &punktfunk_core::input::GamepadSnapshot) {
+        self.buttons = s.buttons;
+        self.left_trigger = s.left_trigger;
+        self.right_trigger = s.right_trigger;
+        self.ls_x = s.ls_x;
+        self.ls_y = s.ls_y;
+        self.rs_x = s.rs_x;
+        self.rs_y = s.rs_y;
+    }
+
     fn frame(&self, index: usize, active_mask: u16) -> crate::gamestream::gamepad::GamepadFrame {
         crate::gamestream::gamepad::GamepadFrame {
             index: index as i16,
@@ -1589,9 +1611,9 @@ impl PadState {
     }
 }
 
-/// Highest pad index addressable on the wire (`flags` field); the uinput manager caps
-/// actual pad creation at its own MAX_PADS.
-const MAX_WIRE_PADS: usize = 16;
+/// Highest pad index addressable on the wire (`flags` field / snapshot `pad`); the uinput
+/// manager caps actual pad creation at its own MAX_PADS.
+const MAX_WIRE_PADS: usize = punktfunk_core::input::MAX_PADS;
 
 /// Backoff between reopen attempts after a host-lifetime service's backend (a capturer) fails
 /// to open or its worker dies, so a persistently-unavailable resource isn't hammered. (The
@@ -1793,6 +1815,9 @@ fn input_thread(
     let mut motion_window = std::time::Instant::now();
     let mut pad_state = [PadState::default(); MAX_WIRE_PADS];
     let mut pad_mask = 0u16;
+    // Last applied snapshot seq per pad (`None` until the first one): the reorder gate for
+    // `InputKind::GamepadState` — a late datagram with an older seq must not roll held state back.
+    let mut pad_seq: [Option<u8>; MAX_WIRE_PADS] = [None; MAX_WIRE_PADS];
     // Rumble is idempotent state on a lossy channel (client-side overflow drops datagrams),
     // so re-send the current state of every rumbling-capable pad every 500 ms — a dropped
     // transition (including a stop) heals on the next refresh.
@@ -1859,6 +1884,32 @@ fn input_thread(
                         pads.handle(&crate::gamestream::gamepad::GamepadEvent::State(frame));
                     }
                 }
+                InputKind::GamepadState => {
+                    // Idempotent full-state snapshot from a capable client (see
+                    // `GamepadSnapshot`): applied only when its seq supersedes the last one, so
+                    // a datagram the network reordered can't roll held state backwards. The
+                    // client refreshes touched pads every ~100 ms, so an unchanged refresh is
+                    // the common case — skip the frame emit then (an XInput packet-number bump
+                    // for identical state is pure churn), but always advance the gate.
+                    use punktfunk_core::input::GamepadSnapshot;
+                    if let Some(snap) = GamepadSnapshot::from_event(&ev) {
+                        let idx = snap.pad as usize;
+                        if idx < MAX_WIRE_PADS && GamepadSnapshot::seq_newer(snap.seq, pad_seq[idx])
+                        {
+                            pad_seq[idx] = Some(snap.seq);
+                            let before = pad_state[idx];
+                            pad_state[idx].set_snapshot(&snap);
+                            let first = pad_mask & (1 << idx) == 0;
+                            if first || pad_state[idx] != before {
+                                pad_mask |= 1 << idx;
+                                let frame = pad_state[idx].frame(idx, pad_mask);
+                                pads.handle(&crate::gamestream::gamepad::GamepadEvent::State(
+                                    frame,
+                                ));
+                            }
+                        }
+                    }
+                }
                 _ => {
                     // Track press/release so a mid-press disconnect can be undone below.
                     match ev.kind {
@@ -2424,9 +2475,25 @@ fn resolve_compositor(
             return Ok(Compositor::Gamescope);
         }
         let available = crate::vdisplay::available();
-        let chosen = pick_compositor(pref, &available, detected).ok_or_else(|| {
-        anyhow!("no usable compositor (no live graphical session for this uid; set PUNKTFUNK_COMPOSITOR or start a desktop/gaming session)")
-    })?;
+        let chosen = match pick_compositor(pref, &available, detected) {
+            Some(c) => c,
+            None => {
+                // No live session — the state a compositor crash leaves behind (gnome-shell
+                // SIGSEGV → GDM greeter, whose auto-login is once-per-boot). If the operator
+                // configured a recovery hook, fire it (debounced) and tell the client to retry:
+                // its next knock lands in the recovered desktop.
+                if crate::vdisplay::try_recover_session() {
+                    anyhow::bail!(
+                        "no live graphical session for this uid — host session recovery launched \
+                         (PUNKTFUNK_RECOVER_SESSION_CMD); retry in a few seconds"
+                    );
+                }
+                anyhow::bail!(
+                    "no usable compositor (no live graphical session for this uid; set \
+                     PUNKTFUNK_COMPOSITOR or start a desktop/gaming session)"
+                );
+            }
+        };
         if !overridden {
             // Point input at the same backend and resolve the gamescope sub-mode (managed where the
             // session infra exists, attach to a foreign gamescope, else per-session bare spawn).
@@ -4299,6 +4366,65 @@ fn build_pipeline(
 mod tests {
     use super::*;
 
+    #[test]
+    fn pad_snapshot_replaces_state_and_seq_gates() {
+        use punktfunk_core::input::{gamepad, GamepadSnapshot};
+        let mut state = PadState::default();
+        let mut last_seq: Option<u8> = None;
+
+        // Legacy accumulation first (an older client), then a snapshot replaces it wholesale.
+        let axis = InputEvent {
+            kind: InputKind::GamepadAxis,
+            _pad: [0; 3],
+            code: gamepad::AXIS_LT,
+            x: 200,
+            y: 0,
+            flags: 0,
+        };
+        assert!(state.apply(&axis));
+        assert_eq!(state.left_trigger, 200);
+
+        let snap = GamepadSnapshot {
+            pad: 0,
+            seq: 1,
+            buttons: gamepad::BTN_A,
+            left_trigger: 255,
+            right_trigger: 0,
+            ls_x: 100,
+            ls_y: -100,
+            rs_x: 0,
+            rs_y: 0,
+        };
+        assert!(GamepadSnapshot::seq_newer(snap.seq, last_seq));
+        last_seq = Some(snap.seq);
+        state.set_snapshot(&snap);
+        assert_eq!(state.left_trigger, 255);
+        assert_eq!(state.buttons, gamepad::BTN_A);
+        assert_eq!((state.ls_x, state.ls_y), (100, -100));
+
+        // A reordered (stale) snapshot must not roll the trigger back.
+        let stale = GamepadSnapshot {
+            seq: 0,
+            left_trigger: 10,
+            ..snap
+        };
+        assert!(!GamepadSnapshot::seq_newer(stale.seq, last_seq));
+
+        // The unchanged-refresh case the input thread skips the frame emit for: identical
+        // payload with a newer seq compares equal after apply.
+        let refresh = GamepadSnapshot { seq: 2, ..snap };
+        assert!(GamepadSnapshot::seq_newer(refresh.seq, last_seq));
+        let before = state;
+        state.set_snapshot(&refresh);
+        assert_eq!(state, before);
+
+        // The snapshot survives the wire roundtrip into the same PadState shape.
+        let dec =
+            GamepadSnapshot::from_event(&InputEvent::decode(&snap.to_event().encode()).unwrap())
+                .unwrap();
+        assert_eq!(dec, snap);
+    }
+
     /// Feed [`RecoveryCadence`] a schedule of event offsets (ms from a common origin) and return
     /// what each `note` produced.
     fn cadence_run(offsets_ms: &[u64]) -> Vec<Option<std::time::Duration>> {

crates/punktfunk-host/src/vdisplay.rs

@@ -712,6 +712,17 @@ pub fn apply_session_env(active: &ActiveSession) {
         Some(sig) => std::env::set_var("HYPRLAND_INSTANCE_SIGNATURE", sig),
         None => std::env::remove_var("HYPRLAND_INSTANCE_SIGNATURE"),
     }
+    // NOTHING live ⇒ every session-scoped var still in the env is a leftover from a previous
+    // connect's retarget, and the availability probes read them: after a gnome-shell crash
+    // (observed 2026-07-10: SIGSEGV → GDM greeter) a stale `XDG_CURRENT_DESKTOP=GNOME` kept
+    // `mutter::is_available()` true, so a client's explicit backend request routed into the dead
+    // session — 45 s create timeouts and a libei error loop instead of the crisp "no live
+    // graphical session" handshake error. Clear them so `available()` reports the truth and the
+    // client fails fast (and, when configured, `try_recover_session` can bring the desktop back).
+    if active.kind == ActiveKind::None {
+        std::env::remove_var("XDG_CURRENT_DESKTOP");
+        std::env::remove_var("WAYLAND_DISPLAY");
+    }
     // Topology (Stage 2): the per-compositor backends (KWin/Mutter) now read
     // [`effective_topology`] directly at create time — the console policy, else the legacy
     // `PUNKTFUNK_{KWIN,MUTTER}_VIRTUAL_PRIMARY` env, else the Auto default (exclusive on the
@@ -721,6 +732,55 @@ pub fn apply_session_env(active: &ActiveSession) {
 #[cfg(not(target_os = "linux"))]
 pub fn apply_session_env(_active: &ActiveSession) {}
 
+/// Fire the operator's session-recovery hook (`PUNKTFUNK_RECOVER_SESSION_CMD`) because a client
+/// connected while NO graphical session is live for this uid — the state a compositor crash
+/// leaves behind (gnome-shell SIGSEGV → GDM greeter, whose auto-login only fires once per boot,
+/// so the box would otherwise sit headless until a walk-up login or a reboot). The command runs
+/// detached via `sh -c` (typically a display-manager restart — see the config docs) and is
+/// debounced to one launch per minute so a retrying client can't stack restarts. Returns whether
+/// a recovery is underway (just launched, or launched within the debounce window), letting the
+/// handshake error tell the client to simply retry.
+#[cfg(target_os = "linux")]
+pub fn try_recover_session() -> bool {
+    let Some(cmd) = crate::config::config().recover_session_cmd.clone() else {
+        return false;
+    };
+    static LAST_LAUNCH: std::sync::Mutex<Option<std::time::Instant>> = std::sync::Mutex::new(None);
+    const DEBOUNCE: std::time::Duration = std::time::Duration::from_secs(60);
+    let mut last = LAST_LAUNCH.lock().unwrap_or_else(|e| e.into_inner());
+    if last.is_some_and(|t| t.elapsed() < DEBOUNCE) {
+        return true; // a launch is already in flight — the retry lands in the recovered session
+    }
+    match std::process::Command::new("/bin/sh")
+        .arg("-c")
+        .arg(&cmd)
+        .stdin(std::process::Stdio::null())
+        .stdout(std::process::Stdio::null())
+        .stderr(std::process::Stdio::null())
+        .spawn()
+    {
+        Ok(mut child) => {
+            *last = Some(std::time::Instant::now());
+            tracing::warn!(cmd = %cmd,
+                "no live graphical session — launched the operator's session-recovery command");
+            // Reap off-thread so the finished child never lingers as a zombie.
+            std::thread::spawn(move || {
+                let _ = child.wait();
+            });
+            true
+        }
+        Err(e) => {
+            tracing::error!(cmd = %cmd, error = %e,
+                "session-recovery command failed to launch");
+            false
+        }
+    }
+}
+#[cfg(not(target_os = "linux"))]
+pub fn try_recover_session() -> bool {
+    false
+}
+
 /// On a **mid-stream** switch to a desktop, the xdg-desktop-portal (D-Bus-activated) and the systemd
 /// `--user` environment can still point at the OLD session, so the host's RemoteDesktop portal opens
 /// against a half-stale env — it accepts events but they don't reach the compositor until a

crates/punktfunk-host/src/vdisplay/linux/mutter.rs

@@ -49,6 +49,18 @@ const APPLY_TEMPORARY: u32 = 1;
 /// Mutter cursor mode: render the cursor into the stream (matches the KWin/gamescope backends).
 const CURSOR_EMBEDDED: u32 = 1;
 
+/// Serializes, process-wide, every Mutter operation that adds/removes a virtual monitor or applies
+/// a monitor configuration. Each of these makes Mutter rebuild its monitor topology, and
+/// *concurrent* rebuilds have segfaulted gnome-shell on-glass twice now: the teardown-side race is
+/// documented at the teardown below, and on 2026-07-10 three simultaneous session setups (three
+/// `RecordVirtual` calls within ~200 µs plus an `ApplyMonitorsConfig`) crashed the shell inside
+/// `meta_monitor_manager_rebuild` — dropping the box to the GDM greeter until a DM restart. One
+/// mutation at a time also keeps [`wait_virtual_connector`] sound: with two virtual outputs
+/// appearing at once, "the connector absent from MY pre-snapshot" can name a sibling's monitor.
+/// Each session runs on its own dedicated thread (see [`session_thread`]), so blocking on a std
+/// mutex — including across the awaits of its single-threaded setup future — is safe.
+static TOPOLOGY_LOCK: std::sync::Mutex<()> = std::sync::Mutex::new(());
+
 /// The Mutter virtual-display driver. Each [`create`](VirtualDisplay::create) spins up a
 /// keepalive thread owning the D-Bus sessions behind the virtual monitor.
 pub struct MutterDisplay {
@@ -146,7 +158,10 @@ impl VirtualDisplay for MutterDisplay {
             })
             .context("spawn Mutter virtual-output thread")?;
 
-        let node_id = match setup_rx.recv_timeout(Duration::from_secs(20)) {
+        // 45 s (was 20 s): setups now queue on TOPOLOGY_LOCK, so a session behind a slow sibling
+        // (whose guard spans up to a ~10 s stream wait + 6 s connector wait + the apply) must
+        // outwait it plus its own handshake before this fires.
+        let node_id = match setup_rx.recv_timeout(Duration::from_secs(45)) {
             Ok(Ok(v)) => v,
             Ok(Err(e)) => bail!("Mutter virtual monitor failed: {e}"),
             Err(_) => bail!("timed out creating the Mutter virtual monitor"),
@@ -181,6 +196,11 @@ impl Drop for StopGuard {
 /// `scale_key`/`remembered_scale` carry the per-client persisted scale: reapplied at connect,
 /// and the user's in-session changes are recorded back under the key (GNOME itself can't — see
 /// [`identity::ScaleMap`](crate::vdisplay::identity)).
+// TOPOLOGY_LOCK is deliberately held across the awaits of the setup/teardown sequences: each
+// session owns this dedicated OS thread and its own single-future runtime, so the guard never
+// blocks a shared executor — it blocks exactly the sibling session threads, which is the point
+// (see TOPOLOGY_LOCK).
+#[allow(clippy::await_holding_lock)]
 fn session_thread(
     setup_tx: Sender<Result<u32, String>>,
     stop: Arc<AtomicBool>,
@@ -201,6 +221,11 @@ fn session_thread(
         }
     };
     rt.block_on(async move {
+        // The whole setup — pre-snapshot → RecordVirtual → ApplyMonitorsConfig — is one
+        // read-modify-write on Mutter's monitor state; hold TOPOLOGY_LOCK across it so concurrent
+        // sessions can't interleave rebuilds (gnome-shell SIGSEGV) or poison each other's
+        // connector diffs. Released before the keepalive park below.
+        let topology_guard = TOPOLOGY_LOCK.lock().unwrap_or_else(|e| e.into_inner());
         // Display-management topology (Stage 2): the console policy's level, resolved to a concrete
         // value. `Extend` leaves the virtual output an extension (no config change); `Primary` makes
         // it the primary monitor but keeps the physicals as secondaries; `Exclusive` makes it the
@@ -288,6 +313,8 @@ fn session_thread(
             }
         }
 
+        drop(topology_guard);
+
         // Park, keeping `session` (and its zbus connection) alive until told to stop. Every ~5 s,
         // read the virtual output's logical-monitor scale and persist a change the user made (GNOME
         // Settings mid-stream) under the client's key — polled rather than teardown-only so a host
@@ -317,6 +344,9 @@ fn session_thread(
         // make_virtual_primary applied an APPLY_TEMPORARY config; Mutter reverts that on its own once
         // the virtual output disappears and our DisplayConfig connection (in `tracked`) closes — so we
         // just drop it here and let the revert happen Mutter-side, never touching the layout ourselves.
+        // The Stop (+ the revert it triggers) is a topology mutation too — take TOPOLOGY_LOCK so a
+        // sibling's teardown or setup can't interleave with the rebuild it causes.
+        let _topology_guard = TOPOLOGY_LOCK.lock().unwrap_or_else(|e| e.into_inner());
         let _ = session.rd_session.call_method("Stop", &()).await;
         drop(tracked);
     });

crates/punktfunk-host/src/vdisplay/windows/manager.rs

@@ -71,7 +71,9 @@ pub(crate) trait VdisplayDriver: Send + Sync {
     unsafe fn open(&self, reap_orphans: bool) -> Result<(OwnedHandle, u32)>;
     /// ADD a virtual monitor at `mode`, pinning the IDD render GPU to `render_luid` first if `Some`, and
     /// requesting `preferred_monitor_id` (the host's per-client stable id; `0` = auto). Returns the REMOVE
-    /// key + target id + the adapter LUID the driver actually used.
+    /// key + target id + the IddCx DISPLAY adapter LUID from the ADD reply
+    /// (`IDARG_OUT_MONITORARRIVAL.OsAdapterLuid` — NOT the render GPU; the driver reports its render
+    /// adapter only in the shared frame header).
     ///
     /// # Safety
     /// `dev` must be the live control handle from [`open`](Self::open).
@@ -100,7 +102,14 @@ pub(crate) trait VdisplayDriver: Send + Sync {
 struct Monitor {
     key: MonitorKey,
     target_id: u32,
+    /// IddCx DISPLAY adapter LUID from the ADD reply (`IDARG_OUT_MONITORARRIVAL.OsAdapterLuid`) —
+    /// NOT the render GPU the driver renders on (the driver reports that one in the shared frame
+    /// header only). Do not compare it against render-GPU picks.
     luid: LUID,
+    /// The render-GPU pin handed to SET_RENDER_ADAPTER at this monitor's ADD (`None` = no GPU was
+    /// selectable). The pin is never re-issued on reuse, so this is what the driver still renders
+    /// on — [`warn_if_pick_moved`] compares the CURRENT pick against it.
+    render_pin: Option<LUID>,
     /// The driver's WUDFHost pid (from the ADD reply) — carried into [`WinCaptureTarget`] so the
     /// IDD-push capturer knows where to duplicate the sealed frame channel's handles.
     wudf_pid: u32,
@@ -475,6 +484,7 @@ impl VirtualDisplayManager {
                 backend = self.driver.name(),
                 "virtual monitor reused (concurrent / reconfigure session)"
             );
+            warn_if_pick_moved(mon);
             return Ok(self.output_for(mon, quit));
         }
 
@@ -487,6 +497,7 @@ impl VirtualDisplayManager {
                     backend = self.driver.name(),
                     "virtual monitor reused (reconnect to a kept monitor)"
                 );
+                warn_if_pick_moved(&mon);
                 if mon.mode != mode {
                     // SAFETY: `reconfigure` needs an exclusive `&mut Monitor` and only touches the live
                     // display topology. `mon` is the local monitor just moved out of the `Lingering`
@@ -562,13 +573,14 @@ impl VirtualDisplayManager {
             crate::vdisplay::policy::Identity::PerClient,
         )
         .unwrap_or(0);
+        let render_pin = resolve_render_pin();
         // SAFETY: `create_monitor`'s own `# Safety` contract guarantees `dev` is the live control
         // handle; we forward it unchanged to `add_monitor`, whose precondition is exactly that.
-        // `resolve_render_pin()` returns an `Option<LUID>` by value (plain `Copy`), so no borrowed
-        // memory crosses the call.
+        // `render_pin` is an `Option<LUID>` by value (plain `Copy`), so no borrowed memory
+        // crosses the call.
         let added = unsafe {
             self.driver
-                .add_monitor(dev, mode, resolve_render_pin(), preferred_id)?
+                .add_monitor(dev, mode, render_pin, preferred_id)?
         };
 
         // Mandatory keepalive: ping inside the watchdog window or the driver tears all displays down.
@@ -724,6 +736,7 @@ impl VirtualDisplayManager {
             key: added.key,
             target_id: added.target_id,
             luid: added.luid,
+            render_pin,
             wudf_pid: added.wudf_pid,
             gdi_name,
             mode,
@@ -1033,6 +1046,36 @@ fn resolve_render_pin() -> Option<LUID> {
     crate::win_adapter::resolve_render_adapter_luid()
 }
 
+/// A reused monitor keeps the render GPU the driver was pinned to at its ADD — the pin is never
+/// re-issued on reuse. When the current pick has moved since then (an operator preference change,
+/// or the max-VRAM tie shifting on identical twin GPUs), say so: the session self-heals (the
+/// IDD-push open rebinds its ring to the driver's actual adapter on a mismatch), but the new pick
+/// only takes effect on the next monitor CREATE, which the mgmt `/display/release` forces.
+/// Compares the pick against the ADD-time PIN — `mon.luid` is the IddCx display adapter and must
+/// not be compared with render-GPU picks.
+fn warn_if_pick_moved(mon: &Monitor) {
+    let Some(pin) = mon.render_pin else { return };
+    let Some(sel) = crate::gpu::selected_gpu() else {
+        return;
+    };
+    let pick = sel.info.luid();
+    if (pick.LowPart, pick.HighPart) != (pin.LowPart, pin.HighPart) {
+        tracing::warn!(
+            pinned_adapter = format!("{:08x}:{:08x}", pin.HighPart, pin.LowPart),
+            current_pick = format!(
+                "{:08x}:{:08x} ({}, {})",
+                pick.HighPart,
+                pick.LowPart,
+                sel.info.name,
+                sel.source.tag()
+            ),
+            "reused virtual monitor is pinned to a different render GPU than the current pick — \
+             the session follows the pinned GPU; free the display (mgmt /display/release) to \
+             recreate it on the picked one"
+        );
+    }
+}
+
 /// A read-only view of the managed monitor for the mgmt `/display/state` endpoint (Goal:
 /// display-management registry facade). Backend-neutral; the [`crate::vdisplay::registry`] facade
 /// maps it into the wire shape.

crates/punktfunk-host/src/vdisplay/windows/pf_vdisplay.rs

@@ -539,17 +539,12 @@ impl VdisplayDriver for PfVdisplayDriver {
                 );
             }
         }
-        if let Some(pin) = render_luid {
-            if luid.LowPart == pin.LowPart && luid.HighPart == pin.HighPart {
-                tracing::info!("pf-vdisplay ADD render adapter matches the pinned GPU (pin took)");
-            } else {
-                tracing::warn!(
-                    add = format!("{:08x}:{:08x}", luid.HighPart, luid.LowPart),
-                    pinned = format!("{:08x}:{:08x}", pin.HighPart, pin.LowPart),
-                    "pf-vdisplay ADD render adapter DIFFERS from pinned — driver ignored SET_RENDER_ADAPTER?"
-                );
-            }
-        }
+        // NOTE: `reply.adapter_luid` is the IddCx DISPLAY adapter
+        // (`IDARG_OUT_MONITORARRIVAL.OsAdapterLuid`), NOT the render GPU, so it can NOT validate
+        // SET_RENDER_ADAPTER — a comparison against the pin here fired "DIFFERS from pinned" on
+        // every ADD (verified on-glass: reply 0x22c05 vs pin 0x15b05 on a single-4090 box). The
+        // driver reports its ACTUAL render adapter in the shared frame header; the IDD-push
+        // capturer checks it there and rebinds on a mismatch.
         Ok(AddedMonitor {
             key: MonitorKey::Session(session_id),
             target_id: reply.target_id,

docs-site/README.md

@@ -4,8 +4,7 @@ The punktfunk documentation site: [Fumadocs](https://fumadocs.dev) on
 [TanStack Start](https://tanstack.com/start) (Vite + Nitro/bun preset).
 
 Content lives in [`content/docs/`](content/docs) as `.md`/`.mdx`. This site is the source of truth
-for the **user-facing** guides; repo-internal design rationale lives in
-[`../design/`](../design/README.md).
+for the **user-facing** guides; design rationale lives in the internal punktfunk-planning repo.
 
 ## API reference
 

docs-site/content/docs/clients.md

@@ -35,9 +35,10 @@ protocol's FEC/encryption extensions, but for a healthy LAN that rarely matters.
 ## Linux desktop client (GTK4)
 
 `punktfunk-client` is the native graphical Linux client — a GTK4 / libadwaita app that speaks
-`punktfunk/1` directly, with hardware decode (VAAPI → dmabuf on Intel/AMD, software fallback),
-PipeWire audio, and SDL3 controllers (rumble, lightbar, DualSense touchpad/motion). Like the Apple
-app it discovers hosts on your network automatically, does PIN pairing, and pins reconnects.
+`punktfunk/1` directly, with hardware decode via Vulkan Video on every GPU vendor (including
+NVIDIA), falling back to VAAPI dmabuf and then software, PipeWire audio, and SDL3 controllers
+(rumble, lightbar, DualSense touchpad/motion). Like the Apple app it discovers hosts on your network
+automatically, does PIN pairing, and pins reconnects.
 
 It ships as a real package, not just a source build — full steps in
 [Install a Client](/docs/install-client#linux-desktop-flatpak):
@@ -73,14 +74,15 @@ The app is in **Google Play Internal Testing** — request a tester invite on ou
 
 `punktfunk-client` for Windows (`clients/windows`) is the native graphical client for Windows — pure
 Rust, the same `punktfunk/1` core as the Apple, Linux, and Android apps, with a **WinUI 3** UI (host
-list, settings, PIN pairing) and the video on a `SwapChainPanel`. It does D3D11VA hardware decode
-(software fallback), 10-bit/HDR present, WASAPI audio + mic, SDL3 controllers (rumble, lightbar,
-DualSense), network discovery, and the full PIN-pairing trust surface. It builds for both `x86_64`
-and `aarch64` and ships as a **signed MSIX**. Launch it and pick a host from the list, just like the
-other native apps.
+list, settings, PIN pairing); the stream itself runs in punktfunk's Vulkan presenter. Its decoder
+tries **Vulkan Video, then D3D11VA, then software**, with 10-bit/HDR present, WASAPI audio + mic,
+SDL3 controllers (rumble, lightbar, DualSense), network discovery, and the full PIN-pairing trust
+surface. It builds for both `x86_64` and `aarch64` and ships as a **signed MSIX**. Launch it and
+pick a host from the list, just like the other native apps.
 
-> The hardware-decode and HDR paths are complete but still pending validation on real GPU hardware.
-> If anything misbehaves, **[Moonlight](/docs/moonlight)** is a proven alternative for Windows.
+> Hardware decode is validated on NVIDIA and Intel GPUs; HDR10 is implemented with on-glass
+> validation still pending. If anything misbehaves, **[Moonlight](/docs/moonlight)** is a proven
+> alternative for Windows.
 
 A headless CLI path exists for scripting/measurement:
 

docs-site/content/docs/configuration.md

@@ -75,6 +75,12 @@ See your desktop page ([KDE](/docs/kde), [GNOME](/docs/gnome)) for when to set t
 | `PUNKTFUNK_KWIN_VIRTUAL_PRIMARY` | `1` | Make the streamed per-session output the sole desktop so plasmashell + windows render on it (not on the headless bootstrap output). Set by the KDE appliance `host.env`. Superseded by the console's **Topology** setting. |
 | `PUNKTFUNK_MUTTER_VIRTUAL_PRIMARY` | `1` | GNOME/Mutter equivalent of the above. |
 
+## Session recovery (Linux)
+
+| Setting | Values | Meaning |
+|---|---|---|
+| `PUNKTFUNK_RECOVER_SESSION_CMD` | command | Operator hook fired (debounced) when a client connects while **no graphical session is live** for the host's user — the state a compositor crash leaves behind (gnome-shell SIGSEGV → GDM greeter, whose auto-login is once-per-boot). Typically `sudo -n systemctl restart gdm` with a matching NOPASSWD sudoers rule, or `systemctl restart display-manager` under a polkit rule; with auto-login enabled the restart brings the desktop back and the client's automatic retry lands in it. Unset/empty = disabled (the default). |
+
 ## Video quality
 
 | Setting | Values | Meaning |
@@ -110,6 +116,13 @@ See your desktop page ([KDE](/docs/kde), [GNOME](/docs/gnome)) for when to set t
 | `PUNKTFUNK_RENDER_ADAPTER` | description substring | Multi-GPU boxes only: force the NVENC/capture GPU by adapter Description substring (e.g. `4090`). Leave unset on single-GPU machines. |
 | `PUNKTFUNK_HOST_CMD` | e.g. `serve --gamestream` | The host subcommand the service launches. Default `serve --gamestream`; use `serve` for a secure native-only host. |
 
+## Network & discovery
+
+| Setting | Values | Meaning |
+|---|---|---|
+| `PUNKTFUNK_MDNS` | `1` · `0` *(default on)* | mDNS adverts (native + GameStream). `0` skips them (same as `--no-mdns`) — for networks/containers where multicast doesn't work; add the host by address in the client instead. |
+| `PUNKTFUNK_DATA_PORT` | port | Pin the per-session video data plane to a fixed UDP port and stream direct (no hole-punch) — open exactly that port in the host firewall. Same as `serve --data-port`; see [Troubleshooting](/docs/troubleshooting). Default: random port + hole-punch. |
+
 ## Auth, API & paths
 
 | Setting | Values | Meaning |

docs-site/content/docs/gnome.md

@@ -66,6 +66,10 @@ journalctl --user -u punktfunk-host -f   # watch it come up and print its identi
 Then bring up [The Web Console](/docs/web-console) to arm pairing and connect a
 [client](/docs/clients). For an always-on box, see the [headless session](#headless-session) below.
 
+Display scaling you set while streaming **sticks per client**: the host remembers each device's
+scale and reapplies it on reconnect — see
+[Persistent scaling](/docs/virtual-displays#persistent-scaling).
+
 ## Headless session
 
 To run with no monitor and no login, keep a GNOME Wayland session up at all times and start the host

docs-site/content/docs/host-cli.md

@@ -35,6 +35,7 @@ punktfunk-host serve --gamestream
 | `--mgmt-bind <IP:PORT>` | Management API address (default `0.0.0.0:47990` — all interfaces, so paired clients can browse the game library over mTLS; pass `127.0.0.1:47990` to keep it loopback-only). |
 | `--mgmt-token <TOKEN>` | Override the bearer token for the management API. |
 | `--no-mdns` | Skip the mDNS adverts (native + GameStream) — for networks/containers where multicast doesn't work. Clients connect via a manually added host instead. Same as `PUNKTFUNK_MDNS=0`. |
+| `--data-port <PORT>` | Pin the per-session video data plane to this fixed UDP port and stream direct (no hole-punch) — open exactly that port in the host firewall. Same as `PUNKTFUNK_DATA_PORT`; default is a random port + hole-punch. |
 
 These are the only flags `serve` accepts.
 
@@ -50,7 +51,8 @@ Every endpoint is documented in the interactive [**API Reference**](/api).
 By default the host **requires pairing** — see [Pairing & Trust](/docs/pairing). On `serve` you
 **arm pairing from the web console** (or mgmt API); the host then displays a 4-digit PIN. Pass `--open` to
 turn off the mandatory-pairing default and serve any device on the network (trusted single-user setups
-only). The pairing flags below are `punktfunk1-host`-only and do **not** apply to `serve`.
+only). `punktfunk1-host` (below) requires pairing by default too; its `--allow-tofu` flag is the
+test-host equivalent of `--open`.
 
 ## `punktfunk1-host`
 
@@ -68,14 +70,16 @@ punktfunk-host punktfunk1-host --source virtual
 | `--seconds <N>` / `--frames <N>` | Bound each session by wall-clock seconds or frame count. |
 | `--max-concurrent <N>` | Stream at most N sessions at once (default 4); overflow waits in the queue. |
 | `--max-sessions <N>` | Exit after N sessions (0 = serve forever). |
-| `--allow-pairing` | Accept PIN pairing; the host prints a PIN when a client pairs. |
-| `--require-pairing` | Only serve paired devices (implies `--allow-pairing`). |
+| `--allow-tofu` | Also accept **unpaired** clients (trust-on-first-use) and advertise pairing as optional. Pairing is required by default; trusted LANs only. (`--allow-pairing`/`--require-pairing` are the old names for the default behaviour and are accepted as no-ops.) |
 | `--pairing-pin <PIN>` | Use a fixed pairing PIN instead of a fresh random one per ceremony. For test harnesses/CI only — a guessable PIN defeats the ceremony's rate limit. |
+| `--data-port <PORT>` | Pin the video data plane to this fixed UDP port and stream direct (no hole-punch). Same as `PUNKTFUNK_DATA_PORT`. |
+| `--idle-timeout-ms <MS>` | Disconnect-detection latency — the QUIC control-connection idle timeout (default 8000). |
 | `--no-mdns` | Skip the `_punktfunk._udp` advert; clients use `--connect HOST:PORT`. Same as `PUNKTFUNK_MDNS=0`. |
 
-`--max-concurrent`, `--allow-pairing`, and `--require-pairing` are **`punktfunk1-host`-only** — `serve` does not
-accept them. On `serve` you arm pairing from the web console instead, and concurrency is fixed at
-the built-in default (4 sessions) rather than settable from the command line.
+`--max-concurrent` and `--allow-tofu` are **`punktfunk1-host`-only** — `serve` does not accept them.
+On `serve` you arm pairing from the web console instead (`--open` is its serve-any-device switch),
+and concurrency is fixed at the built-in default (4 sessions) rather than settable from the command
+line.
 
 Both `serve` and `punktfunk1-host` advertise the host on the network so clients can discover it. List
 hosts from another machine with `punktfunk-probe --discover`. Where multicast doesn't work (some

docs-site/content/docs/install-client.md

@@ -99,9 +99,9 @@ certificate, so you import that certificate once before Windows will install the
 
 3. Launch **Punktfunk** from the Start menu and pick your host.
 
-> The Windows client's hardware-decode (D3D11VA) and HDR paths are complete but still pending
-> validation on real GPU hardware. If anything misbehaves, **[Moonlight](/docs/moonlight)** is a
-> solid alternative for Windows.
+> The Windows client's hardware decode is validated on NVIDIA and Intel GPUs; its HDR path is
+> complete but still pending on-glass validation. If anything misbehaves,
+> **[Moonlight](/docs/moonlight)** is a solid alternative for Windows.
 
 ## macOS
 

docs-site/content/docs/pairing.md

@@ -45,8 +45,9 @@ host's management console, click to arm pairing, and the host displays a 4-digit
 list of paired devices. This works on a headless host over the network — there is no command-line flag
 to arm pairing on `serve`.
 
-(The standalone headless test host, `punktfunk1-host`, takes `--allow-pairing`/`--require-pairing` on its
-command line instead; the production `serve` host arms pairing from the console.)
+(The standalone headless test host, `punktfunk1-host`, requires pairing by default too and takes
+`--allow-tofu` on its command line to accept unpaired clients; the production `serve` host arms
+pairing from the console.)
 
 Then, on the client:
 

docs-site/content/docs/roadmap.md

@@ -16,9 +16,9 @@ see [Status & Progress](/docs/status).
 | GameStream host (works with Moonlight) | ✅ |
 | Native `punktfunk/1` protocol | ✅ |
 | Linux host (KWin · GNOME · gamescope · Sway) | ✅ |
-| Windows host (NVIDIA) | ✅ beta |
+| Windows host (NVIDIA · AMD · Intel) | ✅ beta |
 | Apple client (macOS · iOS · iPadOS · tvOS) | ✅ |
-| Linux client (GTK4) | ✅ |
+| Linux client (GTK4 shell + Vulkan session) | ✅ |
 | Android client (phone · TV) | ✅ |
 | Windows client | 🟡 |
 | Web console + pairing | ✅ |
@@ -45,13 +45,14 @@ see [Status & Progress](/docs/status).
 - **Secure by default** — SPAKE2 PIN pairing with pinned reconnects, one-click delegated approval from
   the web console, and mDNS LAN auto-discovery.
 - **Tuned for latency** — concurrent sessions (stream one desktop to several devices at once),
-  mid-stream resolution renegotiation, a cross-machine clock-skew handshake, a 1 Gbps+ data plane, and
-  an in-app network speed test that informs the bitrate picker.
+  mid-stream resolution renegotiation, a cross-machine clock-skew handshake, a 1 Gbps+ data plane,
+  an in-app network speed test that informs the bitrate picker, and **automatic adaptive bitrate**
+  (the encoder re-targets mid-stream when bitrate is set to Automatic).
 
 ## 🟡 In progress
 
-- **Windows client on-glass validation.** The hardware (D3D11VA) decode, HDR present, and GUI are
-  built and ship as a signed MSIX — they just need verification on real GPU hardware.
+- **Windows client on-glass validation.** Hardware decode and the GUI are validated on NVIDIA and
+  Intel; the HDR present path still needs verification on real HDR hardware.
 - **Apple presenter polish.** The lower-latency `VTDecompressionSession` → `CAMetalLayer` stage-2
   path is now the default; HDR brightness and 4:4:4 still need on-glass validation.
 - **Web console parity.** Surfacing the speed test and bitrate picker the apps already have.

docs-site/content/docs/stats.md

@@ -20,6 +20,22 @@ life:
 4. **displayed** — the picture is handed to the screen (as close to "photons" as the
    platform lets us measure).
 
+## Detail levels
+
+The overlay has four levels — **Off → Compact → Normal → Detailed** — that you cycle live
+in-stream:
+
+| Platform | Cycle with |
+|---|---|
+| Linux · Windows · Steam Deck | **Ctrl+Alt+Shift+S** |
+| macOS / iPad (pointer or trackpad) | **⌃⌥⇧S** or a **three-finger tap** |
+| Android · iPhone | a **three-finger tap** |
+
+**Compact** is a one-line pill (fps · end-to-end ms · Mb/s, plus a loss flag when frames are being
+lost). **Normal** adds the stream line and the p50/p95 headline. **Detailed** adds the decoder path,
+HDR tag, and the per-stage breakdown. You can also set the level a stream starts at in each client's
+Settings. The example below is the **Detailed** view.
+
 ## Reading the overlay
 
 ```
@@ -73,7 +89,7 @@ always spelled out rather than pretending:
 
 | client | headline | why |
 |---|---|---|
-| Windows, macOS/iOS (Metal presenter), Linux | `capture→on-glass` / `capture→displayed` | present instant available (GTK measures at hand-off to the compositor, which adds about one compositor cycle after it) |
+| Windows, macOS/iOS (Metal presenter), Linux | `capture→on-glass` / `capture→displayed` | present instant available (on Linux/Windows, measured right after the Vulkan swapchain present) |
 | Android | `capture→displayed` | MediaCodec's per-frame render callback reports SurfaceFlinger's render timestamp; on the rare window where no callback is delivered (the platform may drop them under load) the HUD falls back to `capture→decoded` |
 | macOS/iOS fallback presenter | `capture→received` | the system video layer hides decode and present timing entirely |
 

docs-site/content/docs/status.md

@@ -15,8 +15,8 @@ A high-level view of where punktfunk stands. The ordered plan of work is on the
 | **Native protocol** — `punktfunk/1` (QUIC control + UDP data, GF(2¹⁶) Leopard FEC + AES-GCM) | ✅ full session planes, validated live |
 | **Windows host** (x64) | 🟡 implemented & shipping as a signed installer; NVIDIA/AMD/Intel encode, newer than the Linux host |
 | **macOS / iOS / iPadOS / tvOS client** | ✅ full client; on-glass-validated stage-2 presenter is the default |
-| **Linux client** (`punktfunk-client`, GTK4/libadwaita) | ✅ full client; VAAPI zero-copy decode + software fallback |
-| **Windows client** (`punktfunk-client`, WinUI 3) | ✅ stage 1 complete; ships as signed MSIX; on-glass hardware validation pending |
+| **Linux client** (`punktfunk-client`, relm4/GTK4 shell + Vulkan session) | ✅ full client; Vulkan Video → VAAPI → software decode |
+| **Windows client** (`punktfunk-client`, WinUI 3) | ✅ ships as signed MSIX; hardware decode validated on NVIDIA + Intel; HDR on-glass validation pending |
 | **Android client** (phone + Android TV) | ✅ full client; hardware HEVC decode + HDR10 |
 | **Web console** (over the management API) | ✅ status · devices · pairing |
 
@@ -54,10 +54,10 @@ host is newer than the Linux host.)
 
 | Client | Highlights |
 |---|---|
-| **macOS / iOS / iPadOS / tvOS** | VideoToolbox HEVC decode, GameController capture, full DualSense feedback, mDNS discovery, PIN pairing + TOFU, network speed test, latency HUD. Stage-2 presenter (`VTDecompressionSession` → `CAMetalLayer`, ~11 ms p50 capture→present) is validated on glass and is the default (stage 1 remains the fallback when Metal is unavailable). Ships as one universal TestFlight build / App Store listing. |
-| **Linux** (`punktfunk-client`) | GTK4/libadwaita. FFmpeg decode with VAAPI → DRM-PRIME dmabuf zero-copy (Intel/AMD; software fallback on NVIDIA), PipeWire audio + mic, SDL3 gamepads incl. DualSense, mDNS discovery, PIN pairing + TOFU, speed test. Ships as Flatpak, apt, rpm, and Arch packages. |
-| **Windows** (`punktfunk-client`) | WinUI 3. D3D11VA zero-copy decode, HDR10, WASAPI audio + mic, SDL3 gamepads incl. DualSense, mDNS discovery, and the full PIN/TOFU trust surface are all implemented. Ships as a signed MSIX (x86_64 + ARM64). **Stage 1 complete; D3D11VA decode, HDR present, and the GUI are pending on-glass validation on real GPU hardware.** |
-| **Android** (phone + Android TV) | Kotlin app with a Rust core over JNI. NDK `AMediaCodec` hardware HEVC decode + HDR10 (Main10/BT.2020 PQ), Opus/Oboe audio + mic, gamepad input with rumble/HID feedback, mDNS discovery, PIN pairing + TOFU (Keystore identity), live stats HUD, and D-pad/controller focus navigation for TV. Ships to the Google Play Internal Testing track. |
+| **macOS / iOS / iPadOS / tvOS** | VideoToolbox HEVC + AV1 decode (AV1 on hardware that decodes it — M3-class Macs, A17 Pro-class iPhones), GameController capture, full DualSense feedback, mDNS discovery, PIN pairing + TOFU, network speed test, latency HUD. Stage-2 presenter (`VTDecompressionSession` → `CAMetalLayer`, ~11 ms p50 capture→present) is validated on glass and is the default (stage 1 remains the fallback when Metal is unavailable). Ships as one universal TestFlight build / App Store listing. |
+| **Linux** (`punktfunk-client`) | relm4/GTK4 shell; streams run in the spawned Vulkan session presenter. Hardware decode via Vulkan Video (all vendors, incl. NVIDIA) with VAAPI dmabuf and software fallbacks, PipeWire audio + mic, SDL3 gamepads incl. DualSense, mDNS discovery, PIN pairing + TOFU, speed test, Skia console UI + tiered stats overlay. Ships as Flatpak, apt, rpm, and Arch packages. |
+| **Windows** (`punktfunk-client`) | WinUI 3 shell; streams run in the Vulkan session presenter with a Vulkan Video → D3D11VA → software decode chain, HDR10, WASAPI audio + mic, SDL3 gamepads incl. DualSense, mDNS discovery, and the full PIN/TOFU trust surface. Ships as a signed MSIX (x86_64 + ARM64). **Hardware decode validated on NVIDIA and Intel; HDR present pending on-glass validation.** |
+| **Android** (phone + Android TV) | Kotlin app with a Rust core over JNI. NDK `AMediaCodec` hardware HEVC decode + HDR10 (Main10/BT.2020 PQ), Opus/Oboe audio + mic, gamepad input with rumble/HID feedback, mDNS discovery, PIN pairing + TOFU (Keystore identity), tiered stats overlay (three-finger tap), custom resolutions, and D-pad/controller focus navigation for TV. Ships to the Google Play Internal Testing track. |
 
 `punktfunk-probe` is a headless reference and measurement client (for testing and
 benchmarking, not everyday use).
@@ -68,7 +68,7 @@ The stack has been validated live on a range of hosts and clients:
 
 - **Hosts:** Ubuntu (GNOME / KDE), Fedora KDE, and Bazzite (gamescope) on machines with
   RTX-class NVIDIA GPUs, across the KWin, gamescope, Mutter, and Sway/wlroots backends.
-- **Clients:** native macOS, Linux, and Android clients against live hosts, plus stock
+- **Clients:** native macOS, Linux, Windows, and Android clients against live hosts, plus stock
   Moonlight clients over the GameStream path.
 - **Cross-machine latency** is measured and skew-corrected (a wall-clock handshake removes
   the inter-machine clock offset), so capture-to-present numbers are valid across the LAN.
@@ -77,11 +77,21 @@ The stack has been validated live on a range of hosts and clients:
 
 Notable capabilities that have landed, newest first:
 
+- **Tiered stats overlay everywhere.** One shared stats vocabulary with **Compact / Normal /
+  Detailed** levels on every client, cycled live in-stream — see
+  [Understanding the Stats Overlay](/docs/stats).
+- **Per-client display scaling on GNOME.** The host persists each client's scale itself and
+  reapplies it on reconnect (GNOME's virtual-monitor API exposes no stable identity to key it on).
+- **Adaptive bitrate.** With bitrate set to **Automatic**, the host re-targets the encoder
+  mid-stream as the link's measured capacity changes.
+- **Gamepad console shell.** The session client carries a full controller-driven shell — host
+  list, PIN pairing, settings, screen transitions, and an on-screen keyboard — usable with no
+  desktop at all (e.g. gamescope on a Steam Deck).
 - **Native Linux client (stage 1).** GTK4/libadwaita app that links `punktfunk-core`
   directly: mDNS host list, TOFU + SPAKE2 PIN pairing, FFmpeg HEVC decode, PipeWire audio
   with mic uplink, SDL3 gamepad capture with rumble/lightbar feedback, layout-independent
-  keyboard, absolute mouse, fullscreen, and a stats overlay. VAAPI → `GdkDmabufTexture`
-  zero-copy decode on Intel/AMD with a proven software fallback.
+  keyboard, absolute mouse, fullscreen, and a stats overlay. (Since re-architected: streams
+  now run in a spawned Vulkan session presenter with Vulkan Video decode on all vendors.)
 - **Delegated pairing approval.** An unpaired device that knocks on a pairing-required host
   appears as a pending request in the web console's Pairing page; one click approves and
   pins its certificate — no PIN fetched out of band.
@@ -112,7 +122,7 @@ See the [Roadmap](/docs/roadmap) for the ordered list. Near-term:
 
 - **True glass-to-glass latency** — combine the client present-stamp (decode → present)
   with the host render → capture term for a complete end-to-end number.
-- **On-glass validation of the Windows client** (D3D11VA decode, HDR present, GUI) on real
-  GPU hardware.
+- **On-glass validation of the Windows client's HDR present path** (hardware decode and the
+  GUI are already validated on NVIDIA and Intel).
 - **gamescope multi-user isolation** — per-session input/audio so concurrent sessions can
   be fully independent desktops (the shared-desktop multi-view case already works).

docs-site/content/docs/virtual-displays.md

@@ -19,9 +19,9 @@ opened on.
 > you also use in person, or a multi-monitor workstation.
 
 > **What's live today:** **keep-alive** (linger, or **forever**), **topology** (extend / primary /
-> exclusive), **conflict handling**, **per-client identity + persistent scaling** (Windows *and*
-> KDE/KWin), and **multi-monitor layout** (several clients as monitors of one desktop) are all
-> enforced. A reconnect always resumes the kept display — even a fast one — instead of spawning a
+> exclusive), **conflict handling**, **per-client identity + persistent scaling** (Windows, KDE/KWin
+> *and* GNOME/Mutter), and **multi-monitor layout** (several clients as monitors of one desktop) are
+> all enforced. A reconnect always resumes the kept display — even a fast one — instead of spawning a
 > second. The remaining gaps are noted inline: the Linux `primary` physical-keep *effect*, Sway
 > `exclusive`, and multi-display for a *single* client (that last is the next stage).
 
@@ -153,7 +153,7 @@ client a *stable display identity*, so your desktop environment keys its per-mon
 |---|---|---|
 | **Windows** | ✅ today | Connect, set scaling in Settings while streaming — Windows remembers it per client. |
 | **KDE / KWin** | ✅ today | Set scaling in System Settings while streaming; KWin keys it to a stable per-client output name and reapplies it on reconnect. Validated live (150 %/125 % survive a full disconnect + reconnect). |
-| **GNOME / Mutter** | ❌ | GNOME's virtual-monitor API exposes no stable identity to key config on. |
+| **GNOME / Mutter** | ✅ today | GNOME's virtual-monitor API exposes no stable identity to key config on, so the **host persists the scale itself**: set scaling in Settings while streaming — the host captures the change, remembers it per client, and reapplies it on reconnect. |
 | **Sway / wlroots** | ❌ | Headless outputs can't carry a stable identity; pin scale in your sway config instead. |
 
 ## Legacy environment knobs

include/punktfunk_core.h

@@ -164,6 +164,10 @@
 // Fixed serialized size of an [`InputEvent`] on the wire (tag + fields).
 #define INPUT_WIRE_LEN (((((1 + 1) + 4) + 4) + 4) + 4)
 
+// The number of gamepads addressable on the wire (`flags` pad index 0..15). Shared by the
+// client's snapshot fold and the host's per-pad accumulators.
+#define MAX_PADS 16
+
 #define PUNKTFUNK_BTN_DPAD_UP 1
 
 #define PUNKTFUNK_BTN_DPAD_DOWN 2
@@ -295,6 +299,15 @@
 #define APP_EXITED_CLOSE_CODE 82
 #endif
 
+#if defined(PUNKTFUNK_FEATURE_QUIC)
+// [`Welcome::host_caps`] bit: the host applies [`InputKind::GamepadState`]
+// (crate::input::InputKind::GamepadState) snapshot events — full per-pad state with a reorder
+// sequence number. A capable client then sends gamepad state as snapshots (idempotent on the
+// lossy datagram plane, periodically refreshed) instead of the fragile per-transition
+// button/axis events; toward a host that doesn't set the bit it keeps the legacy events.
+#define HOST_CAP_GAMEPAD_STATE 1
+#endif
+
 #if defined(PUNKTFUNK_FEATURE_QUIC)
 // [`Hello::video_codecs`] bit: the client can decode H.264 / AVC. The GPU-less **software**
 // encode path (openh264) emits H.264, so a client that wants to stream from a software host MUST
@@ -550,6 +563,17 @@ enum PunktfunkInputKind
     PUNKTFUNK_INPUT_KIND_TOUCH_MOVE = 10,
     // Touch ends. Only `code` (the touch id) is used.
     PUNKTFUNK_INPUT_KIND_TOUCH_UP = 11,
+    // Full gamepad state in one event ([`GamepadSnapshot`]) — idempotent, sequence-numbered.
+    //
+    // The per-transition [`GamepadButton`](Self::GamepadButton)/[`GamepadAxis`](Self::GamepadAxis)
+    // events are fragile on the unreliable datagram plane: a dropped or reordered event corrupts
+    // the host's accumulated pad state until the *next* change (a held trigger stays wrong
+    // indefinitely). A snapshot carries the whole pad, so loss heals on the next send and the
+    // sequence number lets the host drop stale reorders — the same idempotent-state discipline
+    // as the host→client rumble refresh. Sent only when the host advertised
+    // [`HOST_CAP_GAMEPAD_STATE`](crate::quic::HOST_CAP_GAMEPAD_STATE); older hosts keep
+    // receiving the per-transition events.
+    PUNKTFUNK_INPUT_KIND_GAMEPAD_STATE = 12,
 };
 #ifndef __cplusplus
 #if __STDC_VERSION__ >= 202311L

packaging/bazzite/README.md

@@ -339,7 +339,7 @@ control port (UDP 9777) + mDNS + the mgmt/library API (TCP 47990, HTTPS + mTLS).
 host runs `serve --gamestream` (both planes). The
 per-port breakdown below is for reference (or for opening ports by hand); the ports are the code
 constants (`crates/punktfunk-host/src/gamestream/mod.rs`, `mgmt.rs`) and the GameStream-host port-map
-(`design/gamestream-host-plan.md`).
+(punktfunk-planning: `gamestream-host-plan.md`).
 
 **GameStream / Moonlight ports** (fixed; Moonlight derives them from the HTTP base). These only apply
 when the host runs `serve --gamestream` (the bundled unit's default); on a bare-`serve` native-only

packaging/debian/README.md

@@ -2,7 +2,8 @@
 
 `punktfunk-host` is published as a `.deb` to **Gitea's Debian package registry** in the public
 `unom` org, so the Ubuntu hosts update with plain `apt`. CI (`.gitea/workflows/deb.yml`) builds
-and publishes on every push to `main` (a rolling `0.5.0~ciN.g<sha>` build to the **`canary`** apt
+and publishes on every push to `main` (a rolling `<next-minor>~ciN.g<sha>` build — the base is
+derived from the latest stable tag by `scripts/ci/pf-version.sh` — to the **`canary`** apt
 distribution) and on `vX.Y.Z` tags (a clean `X.Y.Z` to the **`stable`** distribution, plus attached
 to the unified Gitea Release). The two are separate apt distributions, so a stable box never jumps
 to a canary build — see [Release Channels](https://punktfunk.unom.io/docs/channels). The repo line

packaging/flatpak/README.md

@@ -4,7 +4,8 @@ The native Linux **client** — the shell (crate `punktfunk-client-linux`, binar
 `punktfunk-client`) plus the Vulkan session binary it execs for streaming (crate
 `punktfunk-client-session`, binary `punktfunk-session`) — is
 published two ways by CI (`.gitea/workflows/flatpak.yml`), on every push to `main` (a rolling
-`0.0.1-ciN.<sha>` build) and on `v*` tags (a clean `X.Y.Z`):
+`<next-minor>-ciN.g<sha>` build, base derived from the latest stable tag by
+`scripts/ci/pf-version.sh`) and on `v*` tags (a clean `X.Y.Z`):
 
 1. **Hosted OSTree repo at `https://flatpak.unom.io`** (recommended) — a GPG-signed Flatpak
    remote served by a static Caddy container on unom-1, so users **install once and then
@@ -58,11 +59,11 @@ per-user (no root, survives SteamOS updates). This is what the Decky plugin uses
 repo above is the better path for a human on the Deck:
 
 ```sh
-# Pick a version: a tag like 1.2.3, or the newest main build's 0.0.1-ciN.gSHA.
+# Pick a version: a tag like 1.2.3, or the newest main build's <next-minor>-ciN.gSHA.
 VER=1.2.3
 URL="https://git.unom.io/api/packages/unom/generic/punktfunk-client-flatpak/$VER/punktfunk-client-$VER.flatpak"
 
-# Flathub must be enabled (it is on the Deck) so the GNOME runtime + ffmpeg-full pull in:
+# Flathub must be enabled (it is on the Deck) so the GNOME runtime + codecs-extra extension pull in:
 flatpak remote-add --user --if-not-exists flathub https://dl.flathub.org/repo/flathub.flatpakrepo
 
 curl -fL -o /tmp/punktfunk-client.flatpak "$URL"
@@ -138,11 +139,19 @@ extension point (auto-downloaded with the runtime; no app-side codec declaration
 `--device=all` (GPU/VAAPI render node + evdev + the hidraw char-devices SDL3 needs for DualSense)
 + `--socket=pulseaudio` (PipeWire-pulse: playback + mic) + `--share=network`. Alongside it:
 `io.unom.Punktfunk.desktop`, `io.unom.Punktfunk.metainfo.xml`, `io.unom.Punktfunk.svg` (all
-installed by the manifest). `cargo-sources.json` (the offline crate cache) is a pure function of
+installed by the manifest). A `vulkan-headers` module supplies what the session binary's ash/Vulkan
+build needs. `cargo-sources.json` (the offline crate cache) is a pure function of
 `Cargo.lock`; CI regenerates it each build and it is **gitignored** — generate it on any box with
 network + `python3`/`aiohttp`/`tomlkit` (`build-flatpak.sh` does this automatically) and, for a
 build host that lacks those (the Deck), rsync the generated file in alongside the manifest.
 
+**Offline Skia:** the session binary's Skia console UI (`pf-console-ui` → `skia-safe`) normally
+downloads prebuilt `libskia` binaries at build time, which is dead in the offline sandbox — so the
+manifest pins a `skia-binaries-….tar.gz` source and points the build at it with
+`SKIA_BINARIES_URL: file://…`. When bumping the `skia-safe`/`skia-bindings` crate version, update
+that pinned tarball (URL + sha256) to the matching `skia-binaries` release or the build breaks
+offline.
+
 ## Hosting the repo (unom-1) + one-time setup
 
 The OSTree repo flatpak-builder produces is GPG-signed in CI and rsynced to unom-1, where a tiny

packaging/linux/steam-deck-gadget/README.md

@@ -12,7 +12,7 @@ binds the kernel `hid-steam` driver, but **Steam Input will not manage it**: Ste
 controller to USB **interface 2**, and a UHID device has no USB interface number (`Interface: -1` in
 Steam's `controller.txt`), so Steam enumerates it but never promotes it. A single-interface DualSense
 is accepted at `-1` (no ambiguity), but the multi-interface Deck specifically needs interface 2. See
-`design/steam-controller-deck-support.md` §11.
+punktfunk-planning: `steam-controller-deck-support.md` §11.
 
 A real multi-interface USB device with the controller on interface 2 requires a **USB gadget**.
 SteamOS ships every piece (`CONFIG_USB_DUMMY_HCD=m`, `CONFIG_USB_RAW_GADGET=m`,

packaging/linux/steam-deck-gadget/usbip-poc/README.md

@@ -17,7 +17,7 @@ sudo modprobe vhci_hcd
 sudo usbip attach -r 127.0.0.1 -b 0-0-0
 ```
 
-See `design/steam-deck-passthrough-plan.md` for the production build plan (vendor-trim the crate to
+See punktfunk-planning: `steam-deck-passthrough-plan.md` for the production build plan (vendor-trim the crate to
 drop the `rusb`/libusb dep; in-process `vhci_hcd` attach to avoid the `usbip` CLI; transport-select
 `raw_gadget`→`usbip`→UHID). `usbip` crate API: a custom `UsbInterfaceHandler` —
 `get_class_specific_descriptor()` = the 9-byte HID descriptor; `handle_urb()` dispatches EP0

packaging/rpm/README.md

@@ -3,7 +3,8 @@
 `punktfunk-host` is published as an RPM to **Gitea's RPM package registry** in the public `unom`
 org (stable groups `bazzite`/`fedora-44`, canary groups `bazzite-canary`/`fedora-44-canary`), so
 Bazzite / Fedora Atomic hosts layer and update it with `rpm-ostree`. CI (`.gitea/workflows/rpm.yml`)
-builds and publishes on every push to `main` (a rolling `0.5.0-0.ciN.g<sha>` build to the `*-canary`
+builds and publishes on every push to `main` (a rolling `<next-minor>-0.ciN.g<sha>` build — the base
+is derived from the latest stable tag by `scripts/ci/pf-version.sh` — to the `*-canary`
 groups) and on `vX.Y.Z` tags (a clean `X.Y.Z-1` to the base groups, plus attached to the unified
 Gitea Release) — separate repos, so a stable box never jumps to a canary build (see
 [Release Channels](https://punktfunk.unom.io/docs/channels)). The `baseurl` below subscribes to the

packaging/windows/README.md

@@ -3,7 +3,7 @@
 A one-file, signed `setup.exe` for the punktfunk streaming **host** on Windows, published to Gitea's
 generic package registry (`punktfunk-host-windows`) by `.gitea/workflows/windows-host.yml`.
 
-> Full picture (drivers-from-source, toolchain, CI, dev loop): **[`design/windows-build-and-packaging.md`](../../design/windows-build-and-packaging.md)**. This README is the `packaging/windows/` file index.
+> Full picture (drivers-from-source, toolchain, CI, dev loop): **punktfunk-planning: `windows-build-and-packaging.md`** (internal planning repo). This README is the `packaging/windows/` file index.
 
 ## Windows 11 22H2+ only (no Windows 10)
 
@@ -135,7 +135,7 @@ fresh install uses the generated random console password — read it from
 > needed). Building from source keeps `.dll`/`.inf`/`.cat` in lockstep. nefcon (the device-node tool -
 > the install creates the `root\pf_vdisplay` node with it, **never** `devgen`, which leaves persistent
 > phantom devices) is fetched + SHA-256-verified from its pinned release in `stage-pf-vdisplay.ps1`. See
-> [`design/windows-build-and-packaging.md`](../../design/windows-build-and-packaging.md) for the toolchain
+> punktfunk-planning: `windows-build-and-packaging.md` (internal planning repo) for the toolchain
 > + signing details.
 
 ## Dev iteration on the test box (driver)
@@ -179,4 +179,5 @@ Push a `vX.Y.Z` tag — one tag releases every platform (see
 [Release Channels](https://punktfunk.unom.io/docs/channels)). The workflow builds, signs, and
 publishes `punktfunk-host-setup-X.Y.Z.exe` + the public `.cer`, refreshes the stable `latest/`
 alias, and attaches the installer to the unified Gitea Release. Main pushes publish rolling
-`0.3.<run>` **canary** builds to the `canary/` alias.
+`<next-minor>.<run>` **canary** builds (base derived from the latest stable tag by
+`scripts/ci/pf-version.ps1`) to the `canary/` alias.

packaging/windows/drivers/pf-dualsense/README.md

@@ -87,7 +87,7 @@ silently breaks them:
   WUDFHost (so the per-instance statics don't collide), and the driver reads its pad index from the
   device Location (`WdfDeviceAllocAndQueryProperty`) to poll its own `*-boot-<index>` bootstrap
   mailbox (the DATA section itself is unnamed — the sealed pad channel,
-  `design/gamepad-channel-sealing.md` — and its `pad_index` is validated against this index on
-  attach).
+  punktfunk-planning: `gamepad-channel-sealing.md` — and its `pad_index` is validated against this
+  index on attach).
 - Port of the WDK `vhidmini2` UMDF2 sample; the DualSense identity + 273-byte descriptor + feature
-  blobs `0x05`/`0x09`/`0x20` come from `crates/punktfunk-host/src/inject/dualsense.rs`.
+  blobs `0x05`/`0x09`/`0x20` come from `crates/punktfunk-host/src/inject/proto/dualsense_proto.rs`.

packaging/windows/drivers/pf-xusb/README.md

@@ -15,7 +15,7 @@ instance (= player slot 0–3) with `CreateFile`, and polls it with buffered IOC
 - registers the XUSB interface with `WdfDeviceCreateDeviceInterface(device, &XUSB_GUID, NULL)`;
 - answers the XUSB IOCTLs (all `METHOD_BUFFERED`, delivered to user mode by the reflector) from
   controller state the host publishes into an **unnamed** shared DATA section reached over the
-  **sealed pad channel** (`design/gamepad-channel-sealing.md`): the host duplicates the section
+  **sealed pad channel** (punktfunk-planning: `gamepad-channel-sealing.md`): the host duplicates the section
   handle into this driver's WUDFHost, bootstrapped via the named `Global\pfxusb-boot-<index>`
   mailbox (`pf_driver_proto::gamepad::PadBootstrap`); a game's rumble (`SET_STATE`) is published
   back for the host to forward to the client.

scripts/ci/check-android-jni-imports.sh

@@ -0,0 +1,74 @@
+#!/bin/sh
+# Post-link floor check for libpunktfunk_android.so: every undefined dynamic symbol must be
+# defined in the NDK's minSdk-level stub libraries, for every built ABI.
+#
+# Why this exists: a Rust cdylib links fine with dangling undefined symbols (lld only rejects them
+# under --no-undefined, which rustc does not pass), so a hard import of an above-floor NDK entry
+# point sails through `cargo ndk --platform 28` and only explodes at `System.loadLibrary` on every
+# device below the symbol's introduction level — 0.9.0 shipped exactly this
+# (AMediaCodec_setOnFrameRenderedCallback, API 33) and bricked the client on all pre-Android-13
+# devices. Above-floor entry points must be dlsym-resolved at runtime instead (see
+# decode::try_set_frame_rate / decode::install_render_callback / adpf). This script makes the
+# violation a build failure. Wired into clients/android/kit/build.gradle.kts after each cargo-ndk
+# build (local + CI).
+#
+# Usage: check-android-jni-imports.sh <ndk-dir> <jniLibs-dir> <api-level>
+set -eu
+
+NDK=$1
+JNILIBS=$2
+API=$3
+
+# Host-agnostic prebuilt dir (linux-x86_64 on CI and the dev box, darwin-* on a Mac).
+PREBUILT=$(echo "$NDK"/toolchains/llvm/prebuilt/*)
+NM="$PREBUILT/bin/llvm-nm"
+[ -x "$NM" ] || { echo "check-android-jni-imports: llvm-nm not found under $NDK" >&2; exit 1; }
+
+triple_for_abi() {
+    case "$1" in
+        arm64-v8a) echo aarch64-linux-android ;;
+        armeabi-v7a) echo arm-linux-androideabi ;;
+        x86_64) echo x86_64-linux-android ;;
+        x86) echo i686-linux-android ;;
+        *) echo "" ;;
+    esac
+}
+
+TMP=$(mktemp -d)
+trap 'rm -rf "$TMP"' EXIT
+
+checked=0
+fail=0
+for so in "$JNILIBS"/*/libpunktfunk_android.so; do
+    [ -f "$so" ] || continue
+    abi=$(basename "$(dirname "$so")")
+    triple=$(triple_for_abi "$abi")
+    if [ -z "$triple" ]; then
+        echo "check-android-jni-imports: unknown ABI dir '$abi' — extend triple_for_abi" >&2
+        exit 1
+    fi
+    stubs="$PREBUILT/sysroot/usr/lib/$triple/$API"
+    [ -d "$stubs" ] || { echo "check-android-jni-imports: no stub dir $stubs" >&2; exit 1; }
+
+    # Strong undefined imports only ('U'; weak 'w'/'v' resolve to null and are fine), and the
+    # stubs' exports, both with the @VERSION / @@VERSION suffix stripped.
+    "$NM" -D --undefined-only "$so" | awk '$1 == "U" { print $2 }' | sed 's/@.*//' | sort -u \
+        > "$TMP/undef"
+    for stub in "$stubs"/*.so; do
+        "$NM" -D --defined-only "$stub" 2>/dev/null | awk '{ print $NF }' | sed 's/@.*//'
+    done | sort -u > "$TMP/defined"
+
+    missing=$(comm -23 "$TMP/undef" "$TMP/defined")
+    if [ -n "$missing" ]; then
+        echo "check-android-jni-imports: $abi libpunktfunk_android.so imports symbols absent from" >&2
+        echo "the API-$API sysroot — System.loadLibrary would fail on devices at the minSdk floor." >&2
+        echo "dlsym-resolve these at runtime instead of hard-linking them:" >&2
+        echo "$missing" | sed 's/^/    /' >&2
+        fail=1
+    fi
+    checked=$((checked + 1))
+done
+
+[ "$checked" -gt 0 ] || { echo "check-android-jni-imports: no libpunktfunk_android.so under $JNILIBS" >&2; exit 1; }
+[ "$fail" -eq 0 ] && echo "check-android-jni-imports: $checked ABI(s) clean at the API-$API floor"
+exit "$fail"

scripts/ci/provision-windows-punktfunk-extras.ps1

@@ -34,13 +34,25 @@ if (Test-Path $rustup) {
 # shipped installer/MSIX stay consistent with punktfunk's MIT OR Apache-2.0 posture.
 # MIGRATION: a runner previously provisioned with the old *gpl-shared* trees must be
 # re-provisioned - delete C:\Users\Public\ffmpeg and C:\Users\Public\ffmpeg-arm64, then re-run.
+# These DLLs are bundled verbatim into the code-signed host installer/MSIX, so the download is
+# SHA-256-pinned (like VB-CABLE below): BtbN's `latest` tag is a ROLLING release whose assets are
+# re-uploaded over time, so an unverified fetch would let a hijacked/MITM'd upstream asset land
+# signed DLLs in users' installs. The pins below were captured 2026-07-10 from the then-current
+# n7.1 lgpl-shared build. When BtbN re-rolls `latest`, this fetch FAILS CLOSED (hash mismatch) —
+# that is intentional: re-download, re-verify the new archive, and update the two pins here.
+#   Refresh a pin:  (Get-FileHash .\ffmpeg-<tag>.zip -Algorithm SHA256).Hash
 function Get-BtbnFfmpeg {
-  param([string]$Dir, [string]$ZipTag)   # ZipTag: 'win64' (x64) or 'winarm64' (ARM64 cross tree)
+  param([string]$Dir, [string]$ZipTag, [string]$Sha)   # ZipTag: 'win64' (x64) or 'winarm64' (ARM64 cross tree)
   if (Test-Path (Join-Path $Dir 'lib\avcodec.lib')) { info "FFmpeg ($ZipTag) already present at $Dir"; return }
-  info "fetching FFmpeg ($ZipTag, BtbN lgpl-shared)"
+  info "fetching FFmpeg ($ZipTag, BtbN lgpl-shared, SHA-256 pinned)"
   $url = "https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-n7.1-latest-$ZipTag-lgpl-shared-7.1.zip"
   $zip = "$Dir.zip"; $tmp = "$Dir-extract"
   Invoke-WebRequest -Uri $url -OutFile $zip -UseBasicParsing
+  $got = (Get-FileHash $zip -Algorithm SHA256).Hash
+  if ($got -ne $Sha) {
+    Remove-Item $zip -Force
+    throw "FFmpeg ($ZipTag) download hash mismatch (got $got, pinned $Sha). BtbN re-rolled the 'latest' build; re-verify the new archive and update the pinned SHA in this script before shipping."
+  }
   if (Test-Path $tmp) { Remove-Item -Recurse -Force $tmp }
   Expand-Archive -Path $zip -DestinationPath $tmp -Force   # BtbN zips have one top-level folder
   $inner = Get-ChildItem $tmp -Directory | Select-Object -First 1
@@ -48,8 +60,8 @@ function Get-BtbnFfmpeg {
   Move-Item -Path $inner.FullName -Destination $Dir
   Remove-Item -Force $zip; Remove-Item -Recurse -Force $tmp -ErrorAction SilentlyContinue
 }
-Get-BtbnFfmpeg -Dir "C:\Users\Public\ffmpeg"       -ZipTag 'win64'
-Get-BtbnFfmpeg -Dir "C:\Users\Public\ffmpeg-arm64" -ZipTag 'winarm64'
+Get-BtbnFfmpeg -Dir "C:\Users\Public\ffmpeg"       -ZipTag 'win64'    -Sha '89F3469706E5D53AEA5CF34AEE63E62CE746E6159D7AEE473D330B02A47558E6'
+Get-BtbnFfmpeg -Dir "C:\Users\Public\ffmpeg-arm64" -ZipTag 'winarm64' -Sha 'D96B4CE08CEBDCC6AD0E3934A3F962915E440EEFB9D73831AFEA4D80E35129A5'
 
 # --- Vulkan-Headers (pf-ffvk's bindgen: libavutil/hwcontext_vulkan.h includes <vulkan/vulkan.h>,
 # and Windows has no system copy). Headers only - the loader (vulkan-1.dll) is a GPU-driver

scripts/ci/retry.sh

@@ -0,0 +1,27 @@
+# shellcheck shell=bash
+# Run a command, retrying with linear backoff:  bash scripts/ci/retry.sh <attempts> <cmd> [args…]
+#
+# WHY THIS EXISTS: the runner box executes many jobs in parallel and its network drops
+# packets under that load — observed as instant "[6] Could not resolve hostname" from
+# flatpak/curl (dropped UDP DNS) and "dial tcp: i/o timeout" from the deploy steps
+# (dropped TCP SYNs to unom-1), each seconds after other network calls in the same job
+# succeeded. Tools with built-in mirror retries (dnf) ride it out; single-shot fetches
+# (flatpak remote-add, rsync, ssh) killed the whole heavy flatpak build on one dropped
+# packet. Wrap every single-shot network command in CI with this instead.
+#
+# Backoff is attempt*10s (10/20/30/…), so 5 attempts spread over ~1.5 min — long enough
+# to outlive a load burst, short enough not to eat the job timeout. The wrapped command's
+# stdout passes through untouched (safe for $(…) capture); retry chatter goes to stderr.
+set -u
+
+attempts="$1"; shift
+rc=1
+for i in $(seq 1 "$attempts"); do
+  "$@" && exit 0
+  rc=$?
+  [ "$i" -eq "$attempts" ] && break
+  echo "::warning::attempt $i/$attempts failed (rc=$rc): $* — retrying in $((i * 10))s" >&2
+  sleep $((i * 10))
+done
+echo "::error::all $attempts attempts failed (rc=$rc): $*" >&2
+exit "$rc"

scripts/host.env.example

@@ -60,6 +60,15 @@ PUNKTFUNK_VIDEO_SOURCE=virtual
                                     # Lower (e.g. 3000) to reclaim kept displays sooner after an
                                     # ungraceful drop; clamped ≥1s, keep-alive ping scales with it so a
                                     # live session never false-disconnects. A deliberate quit is instant.
+# Session recovery hook: fired (debounced, ≥60 s apart) when a client connects while NO graphical
+# session is live for this uid — e.g. a compositor crash dropped the box to the GDM greeter, whose
+# auto-login only runs once per boot, so the box would otherwise need a walk-up login or a reboot.
+# Restarting the display manager re-runs auto-login and the client's retry lands in the recovered
+# desktop. Runs detached via `sh -c` as the host's user, so it needs passwordless privilege for
+# exactly this action (sudoers drop-in: `youruser ALL=(root) NOPASSWD: /usr/bin/systemctl restart
+# gdm`, or a polkit rule + plain `systemctl restart display-manager`). Unset = disabled.
+#PUNKTFUNK_RECOVER_SESSION_CMD=sudo -n systemctl restart gdm
+
 # Full-chroma 4:4:4 (HEVC Range Extensions) — sharper text/desktop, no chroma loss. Honored only on
 # the punktfunk/1 native path when the client advertises 4:4:4 AND the GPU supports it (probed; else
 # the session stays 4:2:0). HEVC-only; independent of 10-bit. NVENC (NVIDIA) is the validated path;

scripts/steamdeck/README.md

@@ -3,7 +3,7 @@
 Run a punktfunk **host** on a Steam Deck — stream its Game Mode (or KDE desktop) *to* other devices.
 (Streaming *to* a Deck is the client; use the Flatpak + [Decky plugin](../../clients/decky/) instead.)
 
-User-facing guide: **docs-site → "Steam Deck (Host)"** (`docs-site/content/docs/steam-deck-host.md`).
+User-facing guide: **docs-site → "SteamOS (Host)"** (`docs-site/content/docs/steamos-host.md`).
 This README is the deep reference for what the scripts do and how to operate them by hand.
 
 ## Why build on-device (not a package or prebuilt binary)

scripts/windows/README.md

@@ -51,7 +51,7 @@ powershell -ExecutionPolicy Bypass -File scripts\windows\build-web.ps1
 ```
 
 `bun install && bun run build` (Nitro `noExternals` -> a self-contained `.output`, no
-`node_modules`/`.npmrc`), then restarts the `PunktfunkWeb` task and checks `:3000/login`. Use
+`node_modules`/`.npmrc`), then restarts the `PunktfunkWeb` task and checks `:47992/login`. Use
 this to iterate on the console against an installed host - `punktfunk-host.exe web setup` (or a
 fresh install) is what creates the task in the first place.
 

tools/latency-probe/README.md

@@ -1,6 +1,6 @@
 # latency-probe
 
-A **glass-to-glass latency** measurement tool (design/implementation-plan §10): it renders a
+A **glass-to-glass latency** measurement tool (punktfunk-planning: `implementation-plan.md` §10): it renders a
 timestamp/QR on the host, reads it back off the client's capture (or a photodiode, for true photons),
 and tracks p50/p99 — so latency regressions are quantifiable rather than felt.
 

tools/loss-harness/README.md

@@ -12,5 +12,5 @@ builds. Use it to sanity-check the FEC before reaching for the real `punktfunk/1
 cargo run -p loss-harness        # from the repo root
 ```
 
-Part of the measurement tooling (design/implementation-plan §10), alongside
+Part of the measurement tooling (punktfunk-planning: `implementation-plan.md` §10), alongside
 [`latency-probe`](../latency-probe/README.md).

web/server/util/auth.ts

@@ -121,11 +121,19 @@ export function isPublicPath(pathname: string): boolean {
 	return false;
 }
 
-/** Validate a post-login redirect target: a same-origin path only. Rejects protocol-
- * relative (`//evil.com`) and absolute URLs to prevent an open redirect. */
+/** Validate a post-login redirect target: a same-origin path only. Resolves `next` against a
+ * sentinel origin and keeps it only if it stays same-origin — rejecting absolute (`https://evil.com`),
+ * protocol-relative (`//evil.com`) AND backslash/tab variants (`/\evil.com`, which the WHATWG URL
+ * parser folds to `//evil.com`) that a plain `startsWith("//")` guard lets through. */
 export function safeNextPath(next: string | undefined): string {
-	if (!next?.startsWith("/") || next.startsWith("//")) return "/";
-	return next;
+	if (!next) return "/";
+	try {
+		const base = "http://pf.invalid";
+		const u = new URL(next, base);
+		return u.origin === base ? u.pathname + u.search + u.hash : "/";
+	} catch {
+		return "/";
+	}
 }
 
 export interface SessionData {