fix(ci/release): make _release_notes_path POSIX-sh safe (deb/decky attach runs under dash)

The v0.19.0 deb + decky release-attach steps source gitea-release.sh under `sh` (dash) and died with "Bad substitution" at _release_notes_path's `${BASH_SOURCE[0]:-$0}` — a bash array subscript dash rejects. So those legs never attached their assets to the release (the bash legs — apple, android, rpm — were fine, which is why the body still seeded and the DMG/.ipa attached). CI always sources this from the repo root, so resolve the notes as docs/releases/<tag>.md relative to CWD and drop $BASH_SOURCE entirely. Verified under dash. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
chore(release): bump workspace version to 0.19.0
2026-07-23 19:03:01 +02:00 · 2026-07-23 18:26:49 +02:00 · 2026-07-23 18:23:38 +02:00 · 2026-07-23 18:18:45 +02:00 · 2026-07-23 17:48:17 +02:00 · 2026-07-23 17:48:17 +02:00
154 changed files with 14795 additions and 2093 deletions
@@ -46,13 +46,23 @@ jobs:
        # SHA-pinned: this workflow's release job carries the signing keystore + Play service-account
        # secrets, so a moved tag on a third-party action could exfiltrate them. v3 = 9fc6c4e.
        uses: android-actions/setup-android@9fc6c4e9069bf8d3d10b2204b1fb8f6ef7065407 # v3
        with:
          # Only platform-tools — NOT the action's default legacy `tools`, whose dependency chain
          # drags in the ~250 MB emulator nobody here runs (instrumentation tests are deferred).
          # That download was the single flakiest piece of this job: the shared runner fleet drops
          # packets under parallel-job load and sdkmanager's streamed unzip turns a truncated
          # stream into "Error on ZipFile unknown archive" (observed 2026-07-22, twice).
          packages: platform-tools
      - name: NDK r30 + platform 36 + build-tools + CMake (libopus cross-build)
        # cmake;3.22.1 installs cmake + ninja under $ANDROID_SDK/cmake/3.22.1/bin — the exact path
        # kit/build.gradle.kts prepends to PATH for cargo-ndk's audiopus_sys (libopus) CMake build.
        # Note: platforms;android-37 is sometimes missing from standard channels; AGP will
        # auto-download it if needed during the build.
-        run: sdkmanager "platform-tools" "platforms;android-36" "build-tools;37.0.0" "ndk;30.0.14904198" "cmake;3.22.1"
+        # retry.sh: sdkmanager is a single-shot multi-hundred-MB fetch, exactly the class the
        # helper exists for (fleet-load packet drops truncate the stream mid-unzip); a failed
        # attempt leaves no partial package behind, so a plain re-invoke is safe.
        run: bash scripts/ci/retry.sh 4 sdkmanager "platform-tools" "platforms;android-36" "build-tools;37.0.0" "ndk;30.0.14904198" "cmake;3.22.1"
      - name: Caches (cargo + gradle)
        uses: actions/cache@v4
@@ -0,0 +1,39 @@
 # Announce a stable release to the Discord #releases channel.
 #
 # This is the deliberate "go" step for a release. Release notes live in the repo at
 # docs/releases/<tag>.md and are seeded into the Gitea release body at creation by the build
 # workflows (scripts/ci/gitea-release.sh), so the release is never noteless. Once every
 # platform's CI is green for a tag, dispatch this workflow with that tag: it re-asserts the notes
 # file over the live release and posts a formatted embed to #releases.
 #
 # Manual on purpose — pressing "go" is the quality gate that says "all platforms built, notes are
 # final, tell the community." It is NOT wired to the tag push, so a half-built or failed release
 # is never announced. Stable-only: a -rc/pre-release tag is refused unless allow_prerelease=true.
 #
 # Requires the repo secret DISCORD_RELEASE_WEBHOOK (the #releases channel webhook URL); GITEA auth
 # reuses REGISTRY_TOKEN like the other release workflows.
 name: announce
 on:
  workflow_dispatch:
    inputs:
      tag:
        description: "Release tag to announce (e.g. v0.18.0)"
        required: true
      allow_prerelease:
        description: "Announce even if the tag is a pre-release (-rc)"
        required: false
        default: "false"
 jobs:
  announce:
    runs-on: ubuntu-24.04
    steps:
      - uses: actions/checkout@v4
      - name: Post release announcement to Discord
        env:
          GITEA_TOKEN: ${{ secrets.REGISTRY_TOKEN }}
          DISCORD_RELEASE_WEBHOOK: ${{ secrets.DISCORD_RELEASE_WEBHOOK }}
          ALLOW_PRERELEASE: ${{ inputs.allow_prerelease }}
        run: bash scripts/ci/discord-announce.sh "${{ inputs.tag }}"
@@ -73,33 +73,20 @@ jobs:
      # sufficient — the Tooling step's dnf install pulls a systemd package upgrade whose RPM
      # trigger re-runs authselect and regenerates this file, undoing the fix. It's reapplied
      # there, right before the first `flatpak` network call.
-      - name: Fix container DNS (drop nss-resolve, resolve over TCP)
+      - name: Fix container DNS (drop nss-resolve)
        run: |
          sed -i 's/resolve \[!UNAVAIL=return\] //' /etc/nsswitch.conf
-          # Resolve over TCP instead of UDP. The documented root cause of the flathub
+          # History: this step used to ALSO force glibc onto TCP DNS (`options use-vc`) because
-          # bootstrap failures (investigated 2026-07-11, see the Tooling step) is this box's
+          # the runner fleet's Docker embedded resolver dropped UDP lookups under parallel-job
-          # Docker embedded resolver at 127.0.0.11 DROPPING UDP lookups while the shared
+          # load (investigated 2026-07-11; v0.15.0/v0.16.0 each burned retry.sh's whole budget).
-          # runner fleet is saturated — a datagram nobody retransmits, so the lookup just
+          # That root cause is now fixed at the infra level (2026-07-22): the runner host runs a
-          # times out. The answer then was to widen retry.sh's budget to 10 attempts (~9 min),
+          # local dnsmasq cache on the docker bridge and daemon.json points every job container
-          # which is enough to outlast a main push's ~8-workflow fan-out but NOT a TAG push's
+          # at it, so lookups terminate on-box instead of crossing the saturated uplink — the
-          # 13: v0.15.0 (twice) and v0.16.0 each burned all 10 attempts and failed the job,
+          # UDP path is reliable again. The TCP path through the same chain proved FLAKY under
-          # each needing a manual re-run.
+          # fleet concurrency (flatpak remote-add failed 10/10 with instant NXDOMAIN while dnf
-          #
+          # in the same container resolved fine), so `use-vc` flipped from mitigation to sole
-          # `use-vc` makes glibc use TCP, where the kernel retransmits and the query cannot be
+          # cause of this leg's failures — removed. retry.sh (10×) stays as the backstop for
-          # silently lost under load. Same resolver, same search path — only the transport
+          # genuine upstream blips.
          # changes, so internal names (git.unom.io) resolve exactly as before; deliberately
          # NO extra nameservers, which would risk answering an internal name from a public
          # resolver. Docker's embedded DNS serves TCP on 127.0.0.11:53 as well as UDP.
          # retry.sh stays as the backstop for genuine upstream blips.
          #
          # Non-fatal: Docker bind-mounts /etc/resolv.conf and can present it read-only, and a
          # DNS tuning that cannot be applied must not be what fails the release build — that
          # would trade an occasional re-run for a hard stop. Falling back to UDP just restores
          # today's behaviour, which retry.sh already covers.
          if ! grep -q '^options .*use-vc' /etc/resolv.conf 2>/dev/null; then
            echo 'options use-vc timeout:3 attempts:3' >> /etc/resolv.conf \
              || echo "::warning::could not set use-vc (read-only resolv.conf?); staying on UDP"
          fi
          cat /etc/resolv.conf || true
      # fedora:43 has no node, but actions/checkout (a JS action) needs it. A plain `run:` step
@@ -393,6 +393,76 @@ jobs:
            -authenticationKeyID "${{ secrets.ASC_API_KEY_ID }}" \
            -authenticationKeyIssuerID "${{ secrets.ASC_API_ISSUER_ID }}"
      - name: iOS — export .ipa (Gitea release + run artifact)
        # The TestFlight step above uploads straight to App Store Connect (destination=upload) and
        # leaves NO .ipa on disk. Re-export the SAME archive with destination=export to get an
        # App Store distribution-signed .ipa for the Gitea release + the run artifacts. Same gate as
        # that archive; a warn+skip (never fails the best-effort iOS leg) if the archive is absent,
        # e.g. a workflow_dispatch with testflight=false. NOTE: an App Store-signed .ipa installs
        # only via TestFlight/App Store, not by direct sideload — it's a release/archival artifact.
        if: gitea.event_name != 'workflow_dispatch' || inputs.testflight == 'true'
        id: ios_ipa
        run: |
          ARCHIVE="$RUNNER_TEMP/Punktfunk-ios.xcarchive"
          if [ ! -d "$ARCHIVE" ]; then
            echo "::warning::iOS archive not found — skipping .ipa export"
            exit 0
          fi
          PROFILE="Punktfunk iOS App Store Distribution"
          WIDGET_PROFILE="Punktfunk iOS Widgets App Store Distribution"
          # destination=export writes the .ipa to -exportPath; otherwise identical manual signing to
          # the upload plist (both profiles, Apple Distribution). No ASC key needed — no network.
          cat > "$RUNNER_TEMP/export-appstore-ipa.plist" <<EOF
          <?xml version="1.0" encoding="UTF-8"?>
          <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
          <plist version="1.0">
          <dict>
            <key>method</key><string>app-store-connect</string>
            <key>destination</key><string>export</string>
            <key>teamID</key><string>$TEAM_ID</string>
            <key>signingStyle</key><string>manual</string>
            <key>signingCertificate</key><string>Apple Distribution</string>
            <key>provisioningProfiles</key>
            <dict>
              <key>io.unom.punktfunk</key><string>$PROFILE</string>
              <key>io.unom.punktfunk.widgets</key><string>$WIDGET_PROFILE</string>
            </dict>
          </dict>
          </plist>
          EOF
          DEVELOPER_DIR="$XCODE_DEV_DIR" xcodebuild -exportArchive \
            -archivePath "$ARCHIVE" \
            -exportOptionsPlist "$RUNNER_TEMP/export-appstore-ipa.plist" \
            -exportPath "$RUNNER_TEMP/export-ipa"
          SRC=$(ls "$RUNNER_TEMP/export-ipa/"*.ipa 2>/dev/null | head -1)
          [ -n "$SRC" ] || { echo "::warning::no .ipa was produced by export"; exit 0; }
          mkdir -p "$GITHUB_WORKSPACE/dist"
          IPA="$GITHUB_WORKSPACE/dist/Punktfunk-$VERSION.ipa"
          mv "$SRC" "$IPA"
          echo "IPA=$IPA" >> "$GITHUB_ENV"
          echo "ipa=dist/Punktfunk-$VERSION.ipa" >> "$GITHUB_OUTPUT"
          echo "exported $IPA"
      - name: Attach .ipa to the workflow run
        if: steps.ios_ipa.outputs.ipa != ''
        # v3, not v4: Gitea's artifact backend identifies as GHES, which upload-artifact@v4 refuses
        # (same reason as android.yml / apple.yml). Download is a zip of the .ipa.
        uses: actions/upload-artifact@v3
        with:
          name: punktfunk-ios-ipa
          path: ${{ steps.ios_ipa.outputs.ipa }}
          if-no-files-found: warn
          retention-days: 30
      - name: Attach .ipa to the Gitea release (stable tags only)
        if: startsWith(gitea.ref, 'refs/tags/v') && steps.ios_ipa.outputs.ipa != ''
        env:
          GITEA_TOKEN: ${{ secrets.REGISTRY_TOKEN }}
        run: |
          . scripts/ci/gitea-release.sh
          RID=$(ensure_release "$GITHUB_REF_NAME" "$GITHUB_REF_NAME" auto)
          upsert_asset "$RID" "$IPA" "Punktfunk-$VERSION.ipa"
      - name: tvOS — archive + upload to TestFlight
        # Canary + stable, the same track as iOS/macOS — the tvOS xcframework slice is now built
        # on every apple push (above), so this matches the iOS step's gate exactly.
@@ -1459,6 +1459,16 @@ dependencies = [
 "version_check",
 ]
 [[package]]
 name = "gethostname"
 version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1bd49230192a3797a9a4d6abe9b3eed6f7fa4c8a8a4947977c6f80025f92cbd8"
 dependencies = [
 "rustix",
 "windows-link 0.2.1 (registry+https://github.com/rust-lang/crates.io-index)",
 ]
 [[package]]
 name = "getrandom"
 version = "0.2.17"
@@ -2184,7 +2194,7 @@ dependencies = [
 [[package]]
 name = "latency-probe"
-version = "0.17.2"
+version = "0.19.0"
 [[package]]
 name = "lazy_static"
@@ -2289,7 +2299,7 @@ dependencies = [
 [[package]]
 name = "libvpl-sys"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "bindgen",
 "cmake",
@@ -2324,7 +2334,7 @@ checksum = "0ceec5bc11778974d1bcb055b18002eba7f4b3518b6a0081b3af5f21666da9ad"
 [[package]]
 name = "loss-harness"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "punktfunk-core",
 ]
@@ -2813,7 +2823,7 @@ checksum = "9b4f627cb1b25917193a259e49bdad08f671f8d9708acfd5fe0a8c1455d87220"
 [[package]]
 name = "pf-capture"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ashpd",
@@ -2829,11 +2839,12 @@ dependencies = [
 "tokio",
 "tracing",
 "windows 0.62.2 (registry+https://github.com/rust-lang/crates.io-index)",
 "x11rb",
 ]
 [[package]]
 name = "pf-client-core"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ash",
@@ -2857,7 +2868,7 @@ dependencies = [
 [[package]]
 name = "pf-clipboard"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ashpd",
@@ -2875,7 +2886,7 @@ dependencies = [
 [[package]]
 name = "pf-console-ui"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ash",
@@ -2896,7 +2907,7 @@ dependencies = [
 [[package]]
 name = "pf-encode"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ash",
@@ -2920,7 +2931,7 @@ dependencies = [
 [[package]]
 name = "pf-ffvk"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "ash",
 "bindgen",
@@ -2929,7 +2940,7 @@ dependencies = [
 [[package]]
 name = "pf-frame"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "libc",
@@ -2941,7 +2952,7 @@ dependencies = [
 [[package]]
 name = "pf-gpu"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "pf-host-config",
@@ -2955,11 +2966,11 @@ dependencies = [
 [[package]]
 name = "pf-host-config"
-version = "0.17.2"
+version = "0.19.0"
 [[package]]
 name = "pf-inject"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ashpd",
@@ -2987,14 +2998,14 @@ dependencies = [
 [[package]]
 name = "pf-paths"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "tracing",
 ]
 [[package]]
 name = "pf-presenter"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ash",
@@ -3009,7 +3020,7 @@ dependencies = [
 [[package]]
 name = "pf-vdisplay"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ashpd",
@@ -3039,7 +3050,7 @@ dependencies = [
 [[package]]
 name = "pf-win-display"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "pf-paths",
@@ -3051,7 +3062,7 @@ dependencies = [
 [[package]]
 name = "pf-zerocopy"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ash",
@@ -3258,7 +3269,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-client-android"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "android_logger",
 "jni",
@@ -3274,7 +3285,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-client-linux"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "async-channel",
@@ -3290,7 +3301,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-client-session"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "pf-client-core",
@@ -3305,7 +3316,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-client-windows"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "async-channel",
 "ffmpeg-next",
@@ -3324,7 +3335,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-core"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "aes-gcm",
 "bytes",
@@ -3356,7 +3367,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-host"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "aes",
 "aes-gcm",
@@ -3440,7 +3451,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-probe"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "mdns-sd",
@@ -3454,7 +3465,7 @@ dependencies = [
 [[package]]
 name = "punktfunk-tray"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "anyhow",
 "ksni",
@@ -3477,7 +3488,7 @@ checksum = "d55d956fa96f5ec02be2e13af0e20391a5aa83d6a074e3ad368959d0fab299ea"
 [[package]]
 name = "pyrowave-sys"
-version = "0.17.2"
+version = "0.19.0"
 dependencies = [
 "bindgen",
 "cmake",
@@ -5886,6 +5897,23 @@ version = "0.6.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1ffae5123b2d3fc086436f8834ae3ab053a283cfac8fe0a0b8eaae044768a4c4"
 [[package]]
 name = "x11rb"
 version = "0.13.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9993aa5be5a26815fe2c3eacfc1fde061fc1a1f094bf1ad2a18bf9c495dd7414"
 dependencies = [
 "gethostname",
 "rustix",
 "x11rb-protocol",
 ]
 [[package]]
 name = "x11rb-protocol"
 version = "0.13.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ea6fc2961e4ef194dcbfe56bb845534d0dc8098940c7e5c012a258bfec6701bd"
 [[package]]
 name = "x509-parser"
 version = "0.16.0"
@@ -48,7 +48,7 @@ exclude = [
 ndk = { path = "clients/android/native/vendor/ndk" }
 [workspace.package]
-version = "0.17.2"
+version = "0.19.0"
 edition = "2021"
 rust-version = "1.82"
 license = "MIT OR Apache-2.0"
@@ -10,7 +10,7 @@
      "name": "MIT OR Apache-2.0",
      "identifier": "MIT OR Apache-2.0"
    },
-    "version": "0.17.2"
+    "version": "0.18.0"
  },
  "paths": {
    "/api/v1/clients": {
@@ -0,0 +1,107 @@
 package io.unom.punktfunk
 import android.content.ClipData
 import android.content.ClipboardManager
 import android.content.Context
 import android.os.Handler
 import android.os.Looper
 import io.unom.punktfunk.kit.NativeBridge
 /**
 * Text clipboard sync for the active session (the desktop-client model, text-only v1):
 *  * **Device → host**: a local copy (the primary-clip listener, plus one probe at start) is
 *    announced as a lazy offer — the text crosses only when the host actually pastes (a
 *    `fetch:` event, answered with the clipboard's current content).
 *  * **Host → device**: a host copy arrives as an `offer:` event and is fetched eagerly into
 *    the system clipboard (Android apps can't lazily materialize a paste from the network
 *    without a content-provider round-trip that isn't worth it here).
 *
 * Loop guard: text set from a host fetch is remembered ([lastFromHost]) so the resulting
 * primary-clip-changed callback doesn't bounce it straight back as a new offer. Clipboard reads
 * happen while the stream is foreground (Android only allows focused-app reads). The native
 * events are drained on a dedicated thread and applied on the main thread; [stop] joins it.
 */
 class ClipboardSync(
    private val context: Context,
    private val handle: Long,
 ) {
    private val main = Handler(Looper.getMainLooper())
    private val cm = context.getSystemService(Context.CLIPBOARD_SERVICE) as ClipboardManager
    @Volatile private var running = true
    private var seq = 0
    private var lastOffered: String? = null
    private var lastFromHost: String? = null
    private var pendingFetch = -1
    private var thread: Thread? = null
    private val clipListener = ClipboardManager.OnPrimaryClipChangedListener { offerLocal() }
    fun start() {
        NativeBridge.nativeClipControl(handle, true)
        cm.addPrimaryClipChangedListener(clipListener)
        thread = Thread({ pollLoop() }, "pf-clipboard").also { it.start() }
        offerLocal() // whatever is already on the clipboard is pasteable host-side right away
    }
    fun stop() {
        running = false
        cm.removePrimaryClipChangedListener(clipListener)
        thread?.join(600) // one poll timeout (250 ms) + slack
        thread = null
    }
    /** Announce the current local text (if it's new and not an echo of a host copy). */
    private fun offerLocal() {
        if (!running) return
        val text = currentClipText() ?: return
        if (text == lastOffered || text == lastFromHost) return
        lastOffered = text
        seq += 1
        NativeBridge.nativeClipOfferText(handle, seq)
    }
    private fun currentClipText(): String? = runCatching {
        cm.primaryClip?.takeIf { it.itemCount > 0 }?.getItemAt(0)
            ?.coerceToText(context)?.toString()?.takeIf { it.isNotEmpty() }
    }.getOrNull()
    private fun pollLoop() {
        while (running) {
            val ev = NativeBridge.nativeNextClip(handle) ?: continue
            if (ev == "closed") return
            main.post { handleEvent(ev) }
        }
    }
    private fun handleEvent(ev: String) {
        if (!running) return
        val parts = ev.split(":", limit = 3)
        when (parts[0]) {
            "offer" -> {
                val offerSeq = parts.getOrNull(1)?.toIntOrNull() ?: return
                if (parts.getOrNull(2) == "1") {
                    pendingFetch = NativeBridge.nativeClipFetchText(handle, offerSeq)
                }
            }
            "fetch" -> {
                val req = parts.getOrNull(1)?.toIntOrNull() ?: return
                val text = currentClipText()
                if (text != null) {
                    NativeBridge.nativeClipServeText(handle, req, text)
                } else {
                    NativeBridge.nativeClipCancel(handle, req)
                }
            }
            "data" -> {
                val xfer = parts.getOrNull(1)?.toIntOrNull() ?: return
                if (xfer != pendingFetch) return // stale/unknown transfer
                pendingFetch = -1
                val text = parts.getOrNull(2)?.takeIf { it.isNotEmpty() } ?: return
                lastFromHost = text
                runCatching { cm.setPrimaryClip(ClipData.newPlainText("Punktfunk", text)) }
            }
            // "state"/"cancel"/"error": nothing to drive in the text-only v1.
        }
    }
 }
@@ -54,6 +54,21 @@ class MainActivity : ComponentActivity() {
    var padKeyProbe: ((KeyEvent) -> Boolean)? = null
    var padMotionProbe: ((MotionEvent) -> Boolean)? = null
    /**
     * Physical-mouse forwarder for the active session (built/released by StreamScreen, like
     * [gamepadRouter]): uncaptured hover/click/wheel forwards as absolute cursor input, captured
     * ([android.view.View.requestPointerCapture]) raw deltas as relative mouse-look. The dispatch
     * overrides below route every SOURCE_MOUSE event here while streaming. Null while not streaming.
     */
    var mouseForwarder: MouseForwarder? = null
    /**
     * TV remote-as-pointer for the active session (StreamScreen builds it on TV devices only):
     * hold SELECT to toggle, then the D-pad glides the host cursor. Consulted first for
     * non-gamepad keys while streaming. Null while not streaming or not a TV.
     */
    var remotePointer: RemotePointer? = null
    /**
     * Set by [StreamScreen] to its disconnect action. The emergency-exit chord (below) invokes it so a
     * couch user with no keyboard/Back can always leave a stream.
@@ -324,9 +339,37 @@ class MainActivity : ComponentActivity() {
                    return true // consumed
                }
            }
            // TV remote-as-pointer sees non-gamepad keys first (SELECT long-press toggles it;
            // while active it owns the D-pad/SELECT/PLAY-PAUSE/BACK).
            if (!event.isFromSource(InputDevice.SOURCE_GAMEPAD)) {
                remotePointer?.let { if (it.onKey(event)) return true }
            }
            // Ctrl+Alt+Shift+Q — the cross-client pointer-capture toggle chord. Swallow both
            // edges of the Q (the modifiers already went over the wire, exactly like desktop).
            if (event.keyCode == KeyEvent.KEYCODE_Q &&
                event.isCtrlPressed && event.isAltPressed && event.isShiftPressed
            ) {
                if (event.action == KeyEvent.ACTION_DOWN && event.repeatCount == 0) {
                    mouseForwarder?.toggleCapture()
                }
                return true
            }
            when (event.keyCode) {
                // A mouse's back/forward buttons already go over the wire as X1/X2 via their
                // BUTTON_* motion edges — but Android ALSO delivers them as key events: the input
                // reader synthesizes KEYCODE_BACK/FORWARD (stamped SOURCE_MOUSE) unconditionally,
                // and a view-level FALLBACK BACK appears when the BUTTON_* press goes unconsumed.
                // Swallow every such duplicate or it doubles as Android navigation and yanks the
                // user out of the stream. A remote/keyboard BACK is never mouse-sourced, so it
                // still falls through to the BackHandler and exits.
                KeyEvent.KEYCODE_BACK, KeyEvent.KEYCODE_FORWARD ->
                    if (event.isFromSource(InputDevice.SOURCE_MOUSE) ||
                        event.flags and KeyEvent.FLAG_FALLBACK != 0
                    ) {
                        return true
                    }
                // Leave these to the system even while streaming.
-                KeyEvent.KEYCODE_BACK, // → BackHandler leaves the stream
+                // (BACK above → BackHandler leaves the stream.)
                KeyEvent.KEYCODE_VOLUME_UP,
                KeyEvent.KEYCODE_VOLUME_DOWN,
                KeyEvent.KEYCODE_VOLUME_MUTE,
@@ -394,6 +437,10 @@ class MainActivity : ComponentActivity() {
    override fun dispatchGenericMotionEvent(event: MotionEvent): Boolean {
        if (streamHandle != 0L) {
            if (gamepadRouter?.onMotion(event) == true) return true
            // Physical mouse (uncaptured): hover motion, wheel, button edges.
            if (event.isFromSource(InputDevice.SOURCE_MOUSE)) {
                mouseForwarder?.let { if (it.onGenericMotion(event)) return true }
            }
            return super.dispatchGenericMotionEvent(event)
        }
        // The Controllers debug screen sees pad motion before the stick→D-pad synthesis below.
@@ -431,6 +478,24 @@ class MainActivity : ComponentActivity() {
        return super.dispatchGenericMotionEvent(event)
    }
    /**
     * Mouse clicks/drags ride the TOUCH stream (the pointer is "down"). While streaming they
     * belong to the mouse forwarder, never to the Compose touch-gesture layer — a physical
     * mouse click must be a real click at the cursor, not a synthesized trackpad tap.
     */
    override fun dispatchTouchEvent(ev: MotionEvent): Boolean {
        if (streamHandle != 0L && ev.isFromSource(InputDevice.SOURCE_MOUSE)) {
            mouseForwarder?.let { if (it.onTouchEvent(ev)) return true }
        }
        return super.dispatchTouchEvent(ev)
    }
    /** The OS is the source of truth for pointer capture (it releases on focus loss). */
    override fun onPointerCaptureChanged(hasCapture: Boolean) {
        super.onPointerCaptureChanged(hasCapture)
        mouseForwarder?.onCaptureChanged(hasCapture)
    }
    /** Keys that drive the console UI — D-pad + face buttons; used to classify the last input source. */
    private fun isConsoleNavKey(kc: Int): Boolean = when (kc) {
        KeyEvent.KEYCODE_DPAD_UP, KeyEvent.KEYCODE_DPAD_DOWN, KeyEvent.KEYCODE_DPAD_LEFT,
@@ -0,0 +1,206 @@
 package io.unom.punktfunk
 import android.view.InputDevice
 import android.view.MotionEvent
 import io.unom.punktfunk.kit.NativeBridge
 import kotlin.math.roundToInt
 /** True when any connected input device is a pointer (USB/BT mouse, or a touchpad driving one). */
 fun hasPhysicalMouse(): Boolean = InputDevice.getDeviceIds().any { id ->
    InputDevice.getDevice(id)?.supportsSource(InputDevice.SOURCE_MOUSE) == true
 }
 /**
 * Physical mouse → wire, in two modes (the iPadOS/desktop model):
 *  * **uncaptured** (default): hover/drag positions forward as absolute cursor moves
 *    (`MouseMoveAbs`, host-normalized against the window size) — desktop-style pointing. The
 *    local cursor is hidden over the stream (StreamScreen sets a TYPE_NULL pointer icon); the
 *    host's own cursor, composited into the video, is the one you see.
 *  * **captured**: the OS pointer is grabbed ([android.view.View.requestPointerCapture]) and raw
 *    relative deltas forward as `MouseMove` — FPS mouse-look. Engaged at stream start / by
 *    clicking into the stream when the "Capture pointer for games" setting is on, and toggled
 *    any time by Ctrl+Alt+Shift+Q (the cross-client chord). Focus loss releases it (the OS
 *    guarantees that); a click re-engages.
 *
 * Buttons ride [MotionEvent.ACTION_BUTTON_PRESS]/RELEASE edges (left/middle/right/back/forward →
 * wire 1/2/3/4/5), the wheel rides [MotionEvent.ACTION_SCROLL] with fractional accumulation so
 * high-resolution wheels don't lose sub-notch travel. Held buttons are tracked and flushed on
 * capture loss / stream exit so nothing sticks on the host. Events reach this class from
 * MainActivity's dispatch overrides (uncaptured) and the capture view's captured-pointer listener.
 */
 class MouseForwarder(
    private val handle: Long,
    private val invertScroll: Boolean,
    private val captureWanted: Boolean,
    private val surfaceSize: () -> Pair<Int, Int>,
 ) {
    /** Capture plumbing, owned by StreamScreen (the focusable capture view). */
    var onRequestCapture: (() -> Unit)? = null
    var onReleaseCapture: (() -> Unit)? = null
    /** Live capture state, updated from [android.app.Activity.onPointerCaptureChanged]. */
    var captured = false
        private set
    /** Chord-released: no auto re-engage (start / click) until the user opts back in. */
    private var userReleased = false
    private val heldButtons = mutableSetOf<Int>()
    private var scrollAccV = 0f
    private var scrollAccH = 0f
    private var moveAccX = 0f
    private var moveAccY = 0f
    /** Uncaptured mouse events on the TOUCH stream (position while a button is down). */
    fun onTouchEvent(ev: MotionEvent): Boolean {
        when (ev.actionMasked) {
            MotionEvent.ACTION_DOWN -> {
                if (captureWanted && !captured && !userReleased) {
                    // The engaging click: grab the pointer and swallow the click (desktop
                    // parity — the click that captures never reaches the host). The paired
                    // BUTTON_RELEASE is dropped by the held-set guard in [button].
                    onRequestCapture?.invoke()
                    return true
                }
                sendAbs(ev)
            }
            MotionEvent.ACTION_MOVE -> sendAbs(ev)
            // Button edges are documented on the generic stream, but be robust to either.
            MotionEvent.ACTION_BUTTON_PRESS -> button(ev.actionButton, true)
            MotionEvent.ACTION_BUTTON_RELEASE -> button(ev.actionButton, false)
        }
        return true
    }
    /** Uncaptured mouse events on the GENERIC stream (hover motion, wheel, button edges). */
    fun onGenericMotion(ev: MotionEvent): Boolean {
        when (ev.actionMasked) {
            MotionEvent.ACTION_HOVER_MOVE -> sendAbs(ev)
            MotionEvent.ACTION_SCROLL -> wheel(ev)
            MotionEvent.ACTION_BUTTON_PRESS -> button(ev.actionButton, true)
            MotionEvent.ACTION_BUTTON_RELEASE -> button(ev.actionButton, false)
            MotionEvent.ACTION_HOVER_ENTER, MotionEvent.ACTION_HOVER_EXIT -> {}
            else -> return false
        }
        return true
    }
    /**
     * Captured-pointer events (the view holds [android.view.View.requestPointerCapture]): x/y ARE
     * the relative deltas ([InputDevice.SOURCE_MOUSE_RELATIVE]), batched samples included. A
     * captured touchpad reports absolute finger coordinates instead — not handled (the touch
     * gesture layer is the touchpad story); returning false leaves those to the framework.
     */
    fun onCapturedPointer(ev: MotionEvent): Boolean {
        if (!ev.isFromSource(InputDevice.SOURCE_MOUSE_RELATIVE)) return false
        when (ev.actionMasked) {
            MotionEvent.ACTION_MOVE -> {
                var dx = 0f
                var dy = 0f
                for (i in 0 until ev.historySize) {
                    dx += ev.getHistoricalX(i)
                    dy += ev.getHistoricalY(i)
                }
                dx += ev.x
                dy += ev.y
                moveAccX += dx
                moveAccY += dy
                val ox = moveAccX.toInt() // truncate toward zero — sub-pixel remainder kept w/ sign
                val oy = moveAccY.toInt()
                if (ox != 0 || oy != 0) {
                    NativeBridge.nativeSendPointerMove(handle, ox, oy)
                    moveAccX -= ox
                    moveAccY -= oy
                }
            }
            MotionEvent.ACTION_BUTTON_PRESS -> button(ev.actionButton, true)
            MotionEvent.ACTION_BUTTON_RELEASE -> button(ev.actionButton, false)
            MotionEvent.ACTION_SCROLL -> wheel(ev)
        }
        return true
    }
    /** Ctrl+Alt+Shift+Q: release the grab, or (re-)engage it — works even when auto-capture is off. */
    fun toggleCapture() {
        if (captured) {
            userReleased = true
            onReleaseCapture?.invoke()
        } else {
            userReleased = false
            onRequestCapture?.invoke()
        }
    }
    /** Auto-engage at stream start (setting on + a mouse actually present). */
    fun engageFromStart() {
        if (captureWanted && !captured && !userReleased && hasPhysicalMouse()) {
            onRequestCapture?.invoke()
        }
    }
    /** From [android.app.Activity.onPointerCaptureChanged] — the OS is the source of truth. */
    fun onCaptureChanged(has: Boolean) {
        captured = has
        // Losing the grab (focus loss, chord) must not leave buttons held on the host.
        if (!has) flushButtons()
    }
    /** Stream teardown: lift anything held and let the grab go. */
    fun release() {
        flushButtons()
        if (captured) onReleaseCapture?.invoke()
    }
    private fun sendAbs(ev: MotionEvent) {
        val (w, h) = surfaceSize()
        if (w <= 0 || h <= 0) return
        NativeBridge.nativeSendPointerAbs(
            handle,
            ev.x.roundToInt().coerceIn(0, w - 1),
            ev.y.roundToInt().coerceIn(0, h - 1),
            w,
            h,
        )
    }
    private fun wheel(ev: MotionEvent) {
        val dir = if (invertScroll) -1f else 1f
        // Android: AXIS_VSCROLL + = up/away, AXIS_HSCROLL + = right — the wire's convention too.
        scrollAccV += ev.getAxisValue(MotionEvent.AXIS_VSCROLL) * 120f * dir
        scrollAccH += ev.getAxisValue(MotionEvent.AXIS_HSCROLL) * 120f * dir
        val v = scrollAccV.toInt()
        if (v != 0) {
            NativeBridge.nativeSendScroll(handle, 0, v)
            scrollAccV -= v
        }
        val h = scrollAccH.toInt()
        if (h != 0) {
            NativeBridge.nativeSendScroll(handle, 1, h)
            scrollAccH -= h
        }
    }
    private fun button(actionButton: Int, down: Boolean) {
        val b = when (actionButton) {
            MotionEvent.BUTTON_PRIMARY -> 1
            MotionEvent.BUTTON_TERTIARY -> 2
            MotionEvent.BUTTON_SECONDARY -> 3
            MotionEvent.BUTTON_BACK -> 4
            MotionEvent.BUTTON_FORWARD -> 5
            else -> return
        }
        if (down) {
            heldButtons.add(b)
            NativeBridge.nativeSendPointerButton(handle, b, true)
        } else if (heldButtons.remove(b)) {
            // Only release what we pressed — drops the release of a swallowed engaging click
            // and anything that raced a capture transition.
            NativeBridge.nativeSendPointerButton(handle, b, false)
        }
    }
    private fun flushButtons() {
        heldButtons.forEach { NativeBridge.nativeSendPointerButton(handle, it, false) }
        heldButtons.clear()
    }
 }
@@ -0,0 +1,193 @@
 package io.unom.punktfunk
 import android.os.Handler
 import android.os.Looper
 import android.view.Choreographer
 import android.view.KeyEvent
 import io.unom.punktfunk.kit.NativeBridge
 import kotlin.math.hypot
 // Hold this long on SELECT (pointer-mode toggle) / PLAY-PAUSE (keyboard toggle) for the long-press
 // action instead of the tap action.
 private const val LONG_PRESS_MS = 800L
 // D-pad glide ballistics, in screen-widths per second: start slow enough to hit a close button,
 // ramp over RAMP_S seconds of continuous hold so crossing the desktop doesn't take all day.
 private const val SPEED_MIN = 0.14f
 private const val SPEED_MAX = 0.70f
 private const val RAMP_S = 1.2f
 /**
 * Android TV remote as a pointer — the Android analogue of the Apple client's Siri-remote pointer,
 * adapted for D-pad-only remotes (most Android TV remotes have no touch surface). For the
 * "TV as a desktop client" use case, where a plain remote is often the only thing in hand.
 *
 * While streaming on a TV, **hold SELECT ≈ 0.8 s** to toggle pointer mode. While active:
 *  * D-pad (held) glides the host cursor with ramping acceleration (relative `MouseMove`,
 *    Choreographer-paced, diagonal-normalized);
 *  * SELECT tap = left click; PLAY/PAUSE tap = right click (Siri-remote parity);
 *  * PLAY/PAUSE held = toggle the on-screen keyboard; BACK = leave pointer mode
 *    (a second BACK then leaves the stream as usual).
 * While inactive, everything except the SELECT long-press passes through untouched (D-pad =
 * arrow keys, SELECT tap = Enter — synthesized on release, since the down was held back to
 * disambiguate the long-press).
 *
 * Only consulted for non-gamepad key events on TV devices (MainActivity gates the calls); all
 * state lives on the main thread.
 */
 class RemotePointer(
    private val handle: Long,
    private val surfaceWidth: () -> Int,
    private val onActiveChanged: (Boolean) -> Unit,
    private val onKeyboardToggle: () -> Unit,
 ) {
    var active = false
        private set
    private val handler = Handler(Looper.getMainLooper())
    private val held = mutableSetOf<Int>() // D-pad keycodes currently down
    private var moveAccX = 0f
    private var moveAccY = 0f
    private var lastFrameNs = 0L
    private var rampSec = 0f
    private var tickerRunning = false
    private var centerLongFired = false
    private var playLongFired = false
    private val centerLong = Runnable {
        centerLongFired = true
        toggle()
    }
    private val playLong = Runnable {
        playLongFired = true
        onKeyboardToggle()
    }
    private val frame = object : Choreographer.FrameCallback {
        override fun doFrame(nowNs: Long) {
            if (!tickerRunning) return
            if (held.isEmpty() || !active) {
                tickerRunning = false
                return
            }
            val dt = if (lastFrameNs == 0L) {
                1f / 60f
            } else {
                ((nowNs - lastFrameNs) / 1e9f).coerceIn(0.001f, 0.1f)
            }
            lastFrameNs = nowNs
            rampSec += dt
            var vx = 0f
            var vy = 0f
            if (KeyEvent.KEYCODE_DPAD_LEFT in held) vx -= 1f
            if (KeyEvent.KEYCODE_DPAD_RIGHT in held) vx += 1f
            if (KeyEvent.KEYCODE_DPAD_UP in held) vy -= 1f
            if (KeyEvent.KEYCODE_DPAD_DOWN in held) vy += 1f
            val mag = hypot(vx, vy)
            if (mag > 0f) {
                val w = surfaceWidth().coerceAtLeast(640)
                val speed = w * (SPEED_MIN + (SPEED_MAX - SPEED_MIN) * (rampSec / RAMP_S).coerceAtMost(1f))
                moveAccX += vx / mag * speed * dt
                moveAccY += vy / mag * speed * dt
                val ox = moveAccX.toInt() // truncate toward zero — sub-pixel remainder kept
                val oy = moveAccY.toInt()
                if (ox != 0 || oy != 0) {
                    NativeBridge.nativeSendPointerMove(handle, ox, oy)
                    moveAccX -= ox
                    moveAccY -= oy
                }
            }
            Choreographer.getInstance().postFrameCallback(this)
        }
    }
    /** One remote key event; true = consumed. Ignore key repeats — the ticker owns motion. */
    fun onKey(event: KeyEvent): Boolean {
        val down = event.action == KeyEvent.ACTION_DOWN
        when (event.keyCode) {
            KeyEvent.KEYCODE_DPAD_CENTER -> {
                if (down) {
                    if (event.repeatCount == 0) {
                        centerLongFired = false
                        handler.postDelayed(centerLong, LONG_PRESS_MS)
                    }
                } else {
                    handler.removeCallbacks(centerLong)
                    if (!centerLongFired) {
                        if (active) {
                            click(1)
                        } else {
                            // The down was held back to disambiguate the long-press, so the
                            // normal path never saw it — synthesize the Enter here instead.
                            NativeBridge.nativeSendKey(handle, 0x0D, true, 0)
                            NativeBridge.nativeSendKey(handle, 0x0D, false, 0)
                        }
                    }
                }
                return true
            }
            KeyEvent.KEYCODE_DPAD_UP, KeyEvent.KEYCODE_DPAD_DOWN,
            KeyEvent.KEYCODE_DPAD_LEFT, KeyEvent.KEYCODE_DPAD_RIGHT,
            -> {
                if (!active) return false
                if (down) {
                    if (held.add(event.keyCode) && held.size == 1) startTicker()
                } else {
                    held.remove(event.keyCode)
                }
                return true
            }
            KeyEvent.KEYCODE_MEDIA_PLAY_PAUSE -> {
                if (!active) return false // inactive: the media-key VK path owns it
                if (down) {
                    if (event.repeatCount == 0) {
                        playLongFired = false
                        handler.postDelayed(playLong, LONG_PRESS_MS)
                    }
                } else {
                    handler.removeCallbacks(playLong)
                    if (!playLongFired) click(3)
                }
                return true
            }
            KeyEvent.KEYCODE_BACK -> {
                if (!active) return false
                if (!down) toggle() // leave pointer mode; the next BACK leaves the stream
                return true
            }
            else -> return false
        }
    }
    /** Stream teardown: stop timers/ticker; nothing wire-held to flush (clicks are edges). */
    fun release() {
        handler.removeCallbacks(centerLong)
        handler.removeCallbacks(playLong)
        active = false
        held.clear()
        tickerRunning = false
    }
    private fun toggle() {
        active = !active
        if (!active) {
            held.clear()
            tickerRunning = false
        }
        onActiveChanged(active)
    }
    private fun startTicker() {
        rampSec = 0f
        lastFrameNs = 0L
        if (!tickerRunning) {
            tickerRunning = true
            Choreographer.getInstance().postFrameCallback(frame)
        }
    }
    private fun click(button: Int) {
        NativeBridge.nativeSendPointerButton(handle, button, true)
        NativeBridge.nativeSendPointerButton(handle, button, false)
    }
 }
@@ -109,6 +109,27 @@ data class Settings(
     * setup where the OS-level pad (lizard mode) is preferred.
     */
    val sc2Capture: Boolean = true,
    /**
     * Lock a physical mouse to the stream ([android.view.View.requestPointerCapture]) and forward
     * raw relative motion — FPS mouse-look, the iPad "Capture pointer for games" twin. Engages at
     * stream start and on a click into the stream; Ctrl+Alt+Shift+Q toggles it live (the chord
     * works even with this off). Off (default): a mouse points absolutely, desktop-style.
     */
    val pointerCapture: Boolean = false,
    /**
     * Flip scroll direction — the mouse wheel and the two-finger touch scroll both. Parity with
     * the Apple/GTK clients' "Invert scroll direction".
     */
    val invertScroll: Boolean = false,
    /**
     * Sync text copied on this device to the host and vice versa while streaming (the desktop
     * clients' shared clipboard, text-only here). Only effective when the host advertises the
     * clipboard capability; the protocol is opt-in per session either way.
     */
    val clipboardSync: Boolean = true,
 )
 /** [Settings.touchMode] values; persisted by name. */
@@ -172,6 +193,9 @@ class SettingsStore(context: Context) {
        autoWakeEnabled = prefs.getBoolean(K_AUTO_WAKE, true),
        rumbleOnPhone = prefs.getBoolean(K_RUMBLE_ON_PHONE, false),
        sc2Capture = prefs.getBoolean(K_SC2_CAPTURE, true),
        pointerCapture = prefs.getBoolean(K_POINTER_CAPTURE, false),
        invertScroll = prefs.getBoolean(K_INVERT_SCROLL, false),
        clipboardSync = prefs.getBoolean(K_CLIPBOARD_SYNC, true),
    )
    fun save(s: Settings) {
@@ -195,6 +219,9 @@ class SettingsStore(context: Context) {
            .putBoolean(K_AUTO_WAKE, s.autoWakeEnabled)
            .putBoolean(K_RUMBLE_ON_PHONE, s.rumbleOnPhone)
            .putBoolean(K_SC2_CAPTURE, s.sc2Capture)
            .putBoolean(K_POINTER_CAPTURE, s.pointerCapture)
            .putBoolean(K_INVERT_SCROLL, s.invertScroll)
            .putBoolean(K_CLIPBOARD_SYNC, s.clipboardSync)
            .apply()
    }
@@ -233,6 +260,9 @@ class SettingsStore(context: Context) {
        const val K_AUTO_WAKE = "auto_wake_enabled"
        const val K_RUMBLE_ON_PHONE = "rumble_on_phone"
        const val K_SC2_CAPTURE = "sc2_capture"
        const val K_POINTER_CAPTURE = "pointer_capture"
        const val K_INVERT_SCROLL = "invert_scroll"
        const val K_CLIPBOARD_SYNC = "clipboard_sync"
        /** Legacy Boolean the enum replaced — read once as the migration default, never written. */
        const val K_TRACKPAD = "trackpad_mode"
@@ -412,6 +412,27 @@ private fun ControlsSettings(s: Settings, update: (Settings) -> Unit, onOpenCont
            style = MaterialTheme.typography.bodySmall,
            color = MaterialTheme.colorScheme.onSurfaceVariant,
        )
        ToggleRow(
            title = "Capture pointer for games",
            subtitle = "Lock a connected mouse to the stream and send raw relative motion " +
                "(mouse-look). Ctrl+Alt+Shift+Q toggles it live; click the stream to re-capture. " +
                "Off: the mouse points at the desktop directly",
            checked = s.pointerCapture,
            onCheckedChange = { on -> update(s.copy(pointerCapture = on)) },
        )
        ToggleRow(
            title = "Invert scroll direction",
            subtitle = "Flip the mouse wheel and two-finger touch scrolling",
            checked = s.invertScroll,
            onCheckedChange = { on -> update(s.copy(invertScroll = on)) },
        )
        ToggleRow(
            title = "Shared clipboard",
            subtitle = "Text copied here pastes on the host and vice versa (hosts with " +
                "clipboard sharing enabled)",
            checked = s.clipboardSync,
            onCheckedChange = { on -> update(s.copy(clipboardSync = on)) },
        )
    }
    SettingsCard {
        SettingDropdown(
@@ -13,6 +13,7 @@ import android.net.wifi.WifiManager
 import android.os.Build
 import android.text.InputType
 import android.util.Log
 import android.view.KeyEvent
 import android.view.SurfaceHolder
 import android.view.SurfaceView
 import android.view.View
@@ -176,6 +177,14 @@ fun StreamScreen(handle: Long, micEnabled: Boolean, onDisconnect: () -> Unit) {
    // "hold to quit" hint overlay. Set from the router's onExitArmed (main thread).
    var exitArming by remember { mutableStateOf(false) }
    // True while the TV remote is acting as a pointer (hold SELECT toggles) — drives the mode hint.
    var remotePointerOn by remember { mutableStateOf(false) }
    // Focus anchor the soft keyboard is summoned onto AND the pointer-capture grab target (a grab
    // needs a focusable view; captured-pointer events land on it). Declared before the effect
    // below so the capture callbacks can reach the view once it exists.
    var keyCapture by remember { mutableStateOf<KeyCaptureView?>(null) }
    DisposableEffect(handle) {
        window?.addFlags(WindowManager.LayoutParams.FLAG_KEEP_SCREEN_ON)
        wifiLocks.forEach { lock ->
@@ -221,6 +230,54 @@ fun StreamScreen(handle: Long, micEnabled: Boolean, onDisconnect: () -> Unit) {
        // Show a "hold to quit" hint the moment the chord completes (the router debounces the actual
        // exit); it clears when the buttons release early or the hold elapses. Runs on the main thread.
        router.onExitArmed = { armed -> exitArming = armed }
        // Physical mouse: uncaptured hover/click/wheel forwards as absolute pointing; captured
        // (setting or the Ctrl+Alt+Shift+Q chord) raw deltas forward as relative mouse-look.
        // The local cursor is hidden over the stream — the host's own cursor, composited into
        // the video, is the one the user sees (twin of the desktop clients' hidden cursor).
        val decor = window?.decorView
        val priorPointerIcon = decor?.pointerIcon
        decor?.pointerIcon = android.view.PointerIcon.getSystemIcon(
            context,
            android.view.PointerIcon.TYPE_NULL,
        )
        val mouse = MouseForwarder(
            handle,
            invertScroll = initialSettings.invertScroll,
            captureWanted = initialSettings.pointerCapture,
            surfaceSize = { (decor?.width ?: 0) to (decor?.height ?: 0) },
        )
        mouse.onRequestCapture = {
            // The grab needs the (focusable) capture view: focus it, then ask. Posted so a
            // request racing view attach/focus settles on the next frame.
            keyCapture?.let { v ->
                v.post {
                    v.requestFocus()
                    v.requestPointerCapture()
                }
            }
        }
        mouse.onReleaseCapture = { keyCapture?.releasePointerCapture() }
        activity?.mouseForwarder = mouse
        // TV remote-as-pointer: hold SELECT ≈ 0.8 s to toggle; the D-pad then glides the host
        // cursor (see RemotePointer). TV only — a phone's remote-less keys stay on the VK path.
        val remote = if (isTv) {
            RemotePointer(
                handle,
                surfaceWidth = { decor?.width ?: 1920 },
                onActiveChanged = { on -> remotePointerOn = on },
                onKeyboardToggle = { keyCapture?.let { it.setImeVisible(!it.imeShown) } },
            )
        } else {
            null
        }
        activity?.remotePointer = remote
        // Shared clipboard (text v1): only when the user setting is on AND the host has a
        // working clipboard service. Protocol-level opt-in + the poll thread live in the sync.
        val clip = if (initialSettings.clipboardSync && NativeBridge.nativeClipSupported(handle)) {
            ClipboardSync(context, handle).also { it.start() }
        } else {
            null
        }
        activity?.setConsoleHighRefreshRate(false) // let the decoder's setFrameRate pick the panel rate
        // Host→client feedback (rumble + DualSense lightbar/LEDs), routed to each controller by pad
        // index via the router; poll threads stopped + joined before the router is released and the
@@ -286,6 +343,7 @@ fun StreamScreen(handle: Long, micEnabled: Boolean, onDisconnect: () -> Unit) {
        }
        onDispose {
            closed.set(true) // from here the handle gets freed; surfaceDestroyed must not touch it
            clip?.stop() // stop + join the clipboard poll thread BEFORE the handle is freed
            feedback.onHidRaw = null
            feedback.stop() // stop + join the poll threads BEFORE the router is released / handle freed
            sc2UsbReceiver?.let { runCatching { context.unregisterReceiver(it) } }
@@ -293,6 +351,12 @@ fun StreamScreen(handle: Long, micEnabled: Boolean, onDisconnect: () -> Unit) {
            router.onExitArmed = null // don't poke Compose state from release()'s disarm while tearing down
            router.release() // flush every slot (nothing sticks host-side) + drop the hot-plug listener
            activity?.gamepadRouter = null
            // Mouse/remote-pointer teardown: lift held buttons, drop the grab, restore the cursor.
            mouse.release()
            activity?.mouseForwarder = null
            remote?.release()
            activity?.remotePointer = null
            decor?.pointerIcon = priorPointerIcon
            activity?.streamHandle = 0L
            activity?.requestStreamExit = null
            // Back in the menus: the SC2 (if present) resumes driving the console UI.
@@ -320,8 +384,12 @@ fun StreamScreen(handle: Long, micEnabled: Boolean, onDisconnect: () -> Unit) {
    // Back gesture = a deliberate exit → signal the quit so the host tears down now (no linger).
    BackHandler { NativeBridge.nativeDisconnectQuit(handle); onDisconnect() }
-    // Focus anchor the three-finger keyboard swipe summons the IME onto (see KeyCaptureView).
+    // Auto-engage pointer capture at stream start (setting on + a mouse actually present).
-    var keyCapture by remember { mutableStateOf<KeyCaptureView?>(null) }
+    // Delayed a beat: the grab needs window focus and the capture view attached.
    LaunchedEffect(handle) {
        delay(400)
        activity?.mouseForwarder?.engageFromStart()
    }
    Box(modifier = Modifier.fillMaxSize()) {
        AndroidView(
@@ -379,23 +447,49 @@ fun StreamScreen(handle: Long, micEnabled: Boolean, onDisconnect: () -> Unit) {
        if (exitArming) {
            ExitChordHint(Modifier.align(Alignment.TopCenter).padding(top = 16.dp))
        }
-        // Invisible 1-px focus anchor for the host-typing soft keyboard (three-finger swipe
+        // Remote-pointer mode hint — the remote's keys are remapped while it's on, so say so.
-        // up in the mouse modes) — it never draws or takes touches, it just owns IME focus.
+        if (remotePointerOn) {
            RemotePointerHint(Modifier.align(Alignment.TopCenter).padding(top = 16.dp))
        }
        // Invisible 1-px focus anchor for the host-typing soft keyboard (three-finger swipe up
        // in the mouse modes) AND the pointer-capture grab target — it never draws or takes
        // touches, it just owns IME focus and receives captured-pointer events.
        AndroidView(
            modifier = Modifier.size(1.dp),
-            factory = { ctx -> KeyCaptureView(ctx).also { keyCapture = it } },
+            factory = { ctx ->
                KeyCaptureView(ctx).also { v ->
                    keyCapture = v
                    // Real IME text path when the host types committed text (see KeyCaptureView).
                    v.textHandle =
                        if (NativeBridge.nativeTextInputSupported(handle)) handle else 0L
                    v.setOnCapturedPointerListener { _, ev ->
                        (ctx as? MainActivity)?.mouseForwarder?.onCapturedPointer(ev) ?: false
                    }
                }
            },
        )
        // Touch input per the Settings model: trackpad/direct-pointer mouse (the shared gesture
        // vocabulary) or real multi-touch passthrough — see TouchInput.kt. Passthrough gets no
        // keyboard gesture: its fingers belong to the host verbatim (a swipe there may BE a
        // host-OS gesture), so intercepting three fingers would corrupt real multi-touch.
        // Stylus lane (design/pen-tablet-input.md §7): against a HOST_CAP_PEN host a stylus
        // splits out of BOTH touch models onto the pen plane; its heartbeat coroutine keeps a
        // stationary held stroke alive (and its cancellation lifts everything on teardown).
        val stylus = remember(handle) {
            if (NativeBridge.nativeHostSupportsPen(handle)) StylusStream(handle) else null
        }
        if (stylus != null) {
            LaunchedEffect(stylus) { stylus.heartbeatLoop() }
        }
        Box(
            Modifier.fillMaxSize().pointerInput(handle, touchMode) {
                when (touchMode) {
-                    TouchMode.TOUCH -> streamTouchPassthrough(handle)
+                    TouchMode.TOUCH -> streamTouchPassthrough(handle, stylus)
                    else -> streamTouchInput(
                        handle,
                        stylus,
                        trackpad = touchMode == TouchMode.TRACKPAD,
                        invertScroll = initialSettings.invertScroll,
                        onCycleStats = { statsVerbosity = statsVerbosity.next() },
                        onKeyboard = { show -> keyCapture?.setImeVisible(show) },
                    )
@@ -423,14 +517,35 @@ private fun ExitChordHint(modifier: Modifier = Modifier) {
    )
 }
 /**
 * The remote-pointer mode cue: while active the remote's keys are remapped (D-pad glides the host
 * cursor, SELECT clicks), so the overlay both confirms the toggle and teaches the vocabulary.
 */
@Composable
 private fun RemotePointerHint(modifier: Modifier = Modifier) {
    Text(
        "Remote pointer — SELECT click · play/pause right-click · hold SELECT to exit",
        modifier = modifier
            .background(Color.Black.copy(alpha = 0.55f), RoundedCornerShape(8.dp))
            .padding(horizontal = 14.dp, vertical = 8.dp),
        color = Color.White,
        fontSize = 15.sp,
    )
 }
 /**
 * Invisible focus anchor for typing on the host: the three-finger swipe summons the device IME
- * onto this view. `TYPE_NULL` puts the IME in "dumb keyboard" mode — it delivers raw [KeyEvent]s
+ * onto this view. Two IME models, picked by the host's capabilities:
- * (no composing text, no autocorrect), which flow through `MainActivity.dispatchKeyEvent` →
+ *  * **Text path** ([textHandle] set — the host advertised `HOST_CAP_TEXT_INPUT`): a real
- * `Keymap.toVk` → the host, the exact path a hardware keyboard takes. Text an IME insists on
+ *    editable [HostTextConnection], so the IME gives autocorrect, gesture typing, non-Latin
- * committing instead still arrives: the non-editable [BaseInputConnection] synthesizes KeyEvents
+ *    composition and emoji, all mirrored to the host as committed text + diffs.
- * for it via `KeyCharacterMap` (with Shift carried as meta state — see the IME-shift wrap in
+ *  * **Fallback** (older host): `TYPE_NULL` puts the IME in "dumb keyboard" mode — raw
- * `MainActivity.dispatchKeyEvent`).
+ *    [KeyEvent]s flow through `MainActivity.dispatchKeyEvent` → `Keymap.toVk` → the host, the
 *    exact path a hardware keyboard takes (with the IME-shift wrap documented there).
 *
 * Doubles as the pointer-capture grab target: a grab needs a focusable view, and captured-pointer
 * events are delivered to it (routed to [MouseForwarder.onCapturedPointer] via the listener the
 * stream screen installs).
 */
 private class KeyCaptureView(context: Context) : View(context) {
    init {
@@ -438,22 +553,171 @@ private class KeyCaptureView(context: Context) : View(context) {
        isFocusableInTouchMode = true
    }
-    override fun onCheckIsTextEditor(): Boolean = true
+    /** The session handle when the host types committed text; `0` = VK-only fallback. */
    var textHandle: Long = 0L
-    override fun onCreateInputConnection(outAttrs: EditorInfo): InputConnection {
+    /** Whether [setImeVisible] last showed the IME — for toggle-style callers (remote pointer). */
    var imeShown = false
        private set
    override fun onCheckIsTextEditor(): Boolean = imeShown
    override fun onCreateInputConnection(outAttrs: EditorInfo): InputConnection? {
        // Only an editor while the user has SUMMONED the keyboard (gesture / remote toggle).
        // This view holds focus for the whole stream (it's the capture anchor), and with an
        // always-live editable connection the IME counts input as active on it — TV IMEs then
        // pop their UI the moment a PHYSICAL keyboard key arrives. With no connection, hardware
        // typing stays on the raw dispatchKeyEvent → Keymap → wire path and no keyboard appears.
        if (!imeShown) return null
        outAttrs.imeOptions = EditorInfo.IME_FLAG_NO_EXTRACT_UI or
            EditorInfo.IME_FLAG_NO_FULLSCREEN or EditorInfo.IME_FLAG_NO_ENTER_ACTION
        return if (textHandle != 0L) {
            outAttrs.inputType = InputType.TYPE_CLASS_TEXT or
                InputType.TYPE_TEXT_FLAG_AUTO_CORRECT or InputType.TYPE_TEXT_FLAG_MULTI_LINE
            HostTextConnection(this, textHandle)
        } else {
            outAttrs.inputType = InputType.TYPE_NULL
-        outAttrs.imeOptions = EditorInfo.IME_FLAG_NO_EXTRACT_UI or EditorInfo.IME_FLAG_NO_FULLSCREEN
+            BaseInputConnection(this, false)
-        return BaseInputConnection(this, false)
+        }
    }
    fun setImeVisible(show: Boolean) {
        val imm = context.getSystemService(Context.INPUT_METHOD_SERVICE) as? InputMethodManager
            ?: return
        imeShown = show
        if (show) {
            requestFocus()
            // The view may already be focused from a null-connection state — restart so the
            // framework re-queries onCreateInputConnection with the gate now open.
            imm.restartInput(this)
            imm.showSoftInput(this, 0)
        } else {
            imm.hideSoftInputFromWindow(windowToken, 0)
            imm.restartInput(this) // gate closed — drop the editable connection
        }
    }
    /**
     * BACK while the summoned keyboard is up: the IME consumes it pre-IME to dismiss itself, so
     * [setImeVisible] never hears about it — sync the gate here or a stale `imeShown` leaves the
     * editable connection live and physical typing re-pops the keyboard.
     */
    override fun onKeyPreIme(keyCode: Int, event: KeyEvent): Boolean {
        if (keyCode == KeyEvent.KEYCODE_BACK && imeShown && event.action == KeyEvent.ACTION_UP) {
            imeShown = false
            (context.getSystemService(Context.INPUT_METHOD_SERVICE) as? InputMethodManager)
                ?.restartInput(this)
        }
        return super.onKeyPreIme(keyCode, event)
    }
 }
 /**
 * IME → host text bridge (the `HOST_CAP_TEXT_INPUT` path): a real **editable** connection, so
 * the IME runs its full machinery (autocorrect, gesture typing, non-Latin composition), mirrored
 * to the host as it happens. The one piece of host-side state tracked is *what the host currently
 * shows of the active composition* ([sentComposition]): composing updates send a common-prefix
 * diff (backspaces + the new suffix) so corrections materialize live on the host; a commit
 * settles it. [setComposingRegion] adopts already-committed text as the active composition
 * (autocorrect-revert / backspace-into-word flows), so the next update diffs against it instead
 * of retyping. Newlines become Enter taps; [deleteSurroundingText] becomes Backspace/Delete taps.
 *
 * Known approximation: diff lengths are counted in Unicode scalars, assuming one host Backspace
 * deletes one scalar — true for the composition text IMEs actually produce (emoji and other
 * multi-unit graphemes commit directly rather than composing).
 */
 private class HostTextConnection(
    view: KeyCaptureView,
    private val handle: Long,
 ) : BaseInputConnection(view, true) {
    /** What the host currently shows of the active composition ("" = none). */
    private var sentComposition = ""
    override fun commitText(text: CharSequence, newCursorPosition: Int): Boolean {
        retype(text.toString())
        sentComposition = ""
        val ok = super.commitText(text, newCursorPosition)
        trimEditable()
        return ok
    }
    override fun setComposingText(text: CharSequence, newCursorPosition: Int): Boolean {
        retype(text.toString())
        return super.setComposingText(text, newCursorPosition)
    }
    override fun finishComposingText(): Boolean {
        // The composition text stands as committed — the host already shows it verbatim.
        sentComposition = ""
        return super.finishComposingText()
    }
    override fun setComposingRegion(start: Int, end: Int): Boolean {
        val e = editable
        if (e != null) {
            val a = start.coerceIn(0, e.length)
            val b = end.coerceIn(0, e.length)
            sentComposition = e.subSequence(minOf(a, b), maxOf(a, b)).toString()
        }
        return super.setComposingRegion(start, end)
    }
    override fun deleteSurroundingText(beforeLength: Int, afterLength: Int): Boolean {
        repeat(beforeLength.coerceIn(0, MAX_TAPS)) { tapVk(VK_BACK) }
        repeat(afterLength.coerceIn(0, MAX_TAPS)) { tapVk(VK_DELETE) }
        return super.deleteSurroundingText(beforeLength, afterLength)
    }
    override fun performEditorAction(actionCode: Int): Boolean {
        tapVk(VK_RETURN)
        return true
    }
    /** Replace the host's view of the composition with [text] via a common-prefix diff. */
    private fun retype(text: String) {
        var common = sentComposition.commonPrefixWith(text)
        // Never split a surrogate pair mid-diff — back off to the pair boundary.
        if (common.isNotEmpty() && common.last().isHighSurrogate()) {
            common = common.dropLast(1)
        }
        val stale = sentComposition.substring(common.length)
        repeat(stale.codePointCount(0, stale.length).coerceAtMost(MAX_TAPS)) { tapVk(VK_BACK) }
        sendText(text.substring(common.length))
        sentComposition = text
    }
    /** Forward literal text, turning newlines into Enter taps (control chars never ride text). */
    private fun sendText(s: String) {
        var chunk = StringBuilder()
        for (ch in s) {
            if (ch == '\n') {
                if (chunk.isNotEmpty()) {
                    NativeBridge.nativeSendText(handle, chunk.toString())
                    chunk = StringBuilder()
                }
                tapVk(VK_RETURN)
            } else {
                chunk.append(ch)
            }
        }
        if (chunk.isNotEmpty()) NativeBridge.nativeSendText(handle, chunk.toString())
    }
    private fun tapVk(vk: Int) {
        NativeBridge.nativeSendKey(handle, vk, true, 0)
        NativeBridge.nativeSendKey(handle, vk, false, 0)
    }
    /** Bound the mirror buffer: once nothing is composing, old text serves no purpose. */
    private fun trimEditable() {
        val e = editable ?: return
        if (getComposingSpanStart(e) == -1 && e.length > 4000) e.clear()
    }
    private companion object {
        const val VK_BACK = 0x08
        const val VK_RETURN = 0x0D
        const val VK_DELETE = 0x2E
        const val MAX_TAPS = 256
    }
 }
@@ -0,0 +1,195 @@
 package io.unom.punktfunk
 import android.view.MotionEvent
 import androidx.compose.ui.ExperimentalComposeUiApi
 import androidx.compose.ui.input.pointer.PointerEvent
 import androidx.compose.ui.input.pointer.PointerType
 import androidx.compose.ui.unit.IntSize
 import io.unom.punktfunk.kit.NativeBridge
 import kotlinx.coroutines.delay
 // Wire PEN_* state bits (punktfunk_core::quic::pen; mirrored, asserted by the Rust shim's docs).
 private const val PEN_IN_RANGE = 1f
 private const val PEN_TOUCHING = 2f
 private const val PEN_BARREL1 = 4f
 private const val PEN_BARREL2 = 8f
 private const val STRIDE = 10
 private const val MAX_SAMPLES = 8
 /**
 * Android stylus → the state-full pen plane (design/pen-tablet-input.md §7): pressure, tilt
 * (`AXIS_TILT`, radians from the surface normal), azimuth (`AXIS_ORIENTATION` — Android's 0 =
 * "pointed away from the user" IS the wire's north, no offset needed), hover with
 * `AXIS_DISTANCE`, the eraser tool, both stylus barrel buttons, and historical (coalesced)
 * samples batched oldest-first for full capture-rate fidelity. Android has no barrel-roll
 * axis — roll stays unknown on this client.
 *
 * Both touch loops call [intercept] first; stylus/eraser pointers are consumed here (against a
 * pen-capable host) and never reach the finger paths, independent of the touch-input mode.
 * [heartbeatLoop] implements the ≤100 ms keepalive wire contract: a stationary held stylus is
 * silent in Android's input pipeline, and the host force-releases a stroke after 200 ms
 * without samples.
 */
 internal class StylusStream(private val handle: Long) {
    private var inRange = false
    private var touching = false
    private var sawHover = false
    private val last = FloatArray(STRIDE)
    private val batch = FloatArray(MAX_SAMPLES * STRIDE)
    init {
        idle(last)
    }
    /**
     * Consume the event's stylus pointers into pen samples. Returns true when this event
     * carried any (the caller's finger/gesture handling must then skip those changes).
     */
    @OptIn(ExperimentalComposeUiApi::class)
    fun intercept(ev: PointerEvent, size: IntSize): Boolean {
        val stylusChanges = ev.changes.filter {
            it.type == PointerType.Stylus || it.type == PointerType.Eraser
        }
        if (stylusChanges.isEmpty()) return false
        stylusChanges.forEach { it.consume() }
        val me = ev.motionEvent ?: return true
        if (size.width <= 0 || size.height <= 0) return true
        // At most one stylus exists — find its pointer index by tool type.
        val idx = (0 until me.pointerCount).firstOrNull {
            me.getToolType(it) == MotionEvent.TOOL_TYPE_STYLUS ||
                me.getToolType(it) == MotionEvent.TOOL_TYPE_ERASER
        } ?: return true
        when (me.actionMasked) {
            MotionEvent.ACTION_DOWN, MotionEvent.ACTION_POINTER_DOWN,
            MotionEvent.ACTION_MOVE,
            -> {
                touching = true
                inRange = true
                emitSamples(me, idx, size)
            }
            MotionEvent.ACTION_HOVER_ENTER, MotionEvent.ACTION_HOVER_MOVE -> {
                sawHover = true
                inRange = true
                touching = false
                emitSamples(me, idx, size)
            }
            MotionEvent.ACTION_UP, MotionEvent.ACTION_POINTER_UP -> {
                touching = false
                // Hover-capable hardware keeps proximity (HOVER_EXIT owns the leave);
                // anything else leaves range on lift — the host never parks a phantom pen.
                inRange = sawHover
                emitSamples(me, idx, size)
            }
            MotionEvent.ACTION_HOVER_EXIT, MotionEvent.ACTION_CANCEL -> release()
            else -> {}
        }
        return true
    }
    /** Session/composition teardown: leave range so the host lifts anything still inked. */
    fun reset() {
        if (inRange || touching) release()
        sawHover = false
    }
    /** The ≤100 ms keepalive (80 ms leaves headroom for one lost datagram). Runs until
     *  cancelled; resends the last state-full sample while the pen is in range. */
    suspend fun heartbeatLoop() {
        try {
            while (true) {
                delay(80)
                if (inRange || touching) {
                    last[9] = 0f // dt
                    NativeBridge.nativeSendPen(handle, last, 1)
                }
            }
        } finally {
            reset()
        }
    }
    private fun release() {
        touching = false
        inRange = false
        last[0] = 0f // state: out of range
        last[4] = 0f // pressure
        NativeBridge.nativeSendPen(handle, last, 1)
    }
    /** Historical (coalesced) samples oldest-first, then the current one — a single batch. */
    private fun emitSamples(me: MotionEvent, idx: Int, size: IntSize) {
        val history = minOf(me.historySize, MAX_SAMPLES - 1)
        var count = 0
        var prevT = if (history > 0) me.getHistoricalEventTime(0) else me.eventTime
        for (h in (me.historySize - history) until me.historySize) {
            val t = me.getHistoricalEventTime(h)
            fill(
                batch, count * STRIDE, size,
                x = me.getHistoricalX(idx, h), y = me.getHistoricalY(idx, h),
                pressure = me.getHistoricalPressure(idx, h),
                tiltRad = me.getHistoricalAxisValue(MotionEvent.AXIS_TILT, idx, h),
                orientRad = me.getHistoricalAxisValue(MotionEvent.AXIS_ORIENTATION, idx, h),
                distance = me.getHistoricalAxisValue(MotionEvent.AXIS_DISTANCE, idx, h),
                buttons = me.buttonState, tool = me.getToolType(idx),
                dtUs = ((t - prevT) * 1000).coerceIn(0, 65535).toFloat(),
            )
            prevT = t
            count++
        }
        fill(
            batch, count * STRIDE, size,
            x = me.getX(idx), y = me.getY(idx), pressure = me.getPressure(idx),
            tiltRad = me.getAxisValue(MotionEvent.AXIS_TILT, idx),
            orientRad = me.getAxisValue(MotionEvent.AXIS_ORIENTATION, idx),
            distance = me.getAxisValue(MotionEvent.AXIS_DISTANCE, idx),
            buttons = me.buttonState, tool = me.getToolType(idx),
            dtUs = ((me.eventTime - prevT) * 1000).coerceIn(0, 65535).toFloat(),
        )
        count++
        batch.copyInto(last, 0, (count - 1) * STRIDE, count * STRIDE)
        NativeBridge.nativeSendPen(handle, batch, count)
    }
    private fun fill(
        out: FloatArray,
        off: Int,
        size: IntSize,
        x: Float,
        y: Float,
        pressure: Float,
        tiltRad: Float,
        orientRad: Float,
        distance: Float,
        buttons: Int,
        tool: Int,
        dtUs: Float,
    ) {
        var state = 0f
        if (inRange || touching) state += PEN_IN_RANGE
        if (touching) state += PEN_TOUCHING
        if (buttons and MotionEvent.BUTTON_STYLUS_PRIMARY != 0) state += PEN_BARREL1
        if (buttons and MotionEvent.BUTTON_STYLUS_SECONDARY != 0) state += PEN_BARREL2
        out[off + 0] = state
        out[off + 1] = if (tool == MotionEvent.TOOL_TYPE_ERASER) 1f else 0f
        out[off + 2] = (x / (size.width - 1).coerceAtLeast(1)).coerceIn(0f, 1f)
        out[off + 3] = (y / (size.height - 1).coerceAtLeast(1)).coerceIn(0f, 1f)
        out[off + 4] = if (touching) pressure.coerceIn(0f, 1f) else 0f
        // AXIS_DISTANCE units are device-arbitrary; 0..1 covers real hardware, and 0 while
        // hovering legitimately means "at the hover floor".
        out[off + 5] = if (touching) 0f else distance.coerceIn(0f, 1f)
        out[off + 6] = Math.toDegrees(tiltRad.toDouble()).toFloat().coerceIn(0f, 90f)
        // AXIS_ORIENTATION: 0 = pointed away from the user (= wire north), clockwise, −π..π.
        out[off + 7] = ((Math.toDegrees(orientRad.toDouble()) + 360.0) % 360.0).toFloat()
        out[off + 8] = -1f // no barrel-roll axis on Android
        out[off + 9] = dtUs
    }
    private fun idle(out: FloatArray) {
        out.fill(0f)
        out[5] = -1f // distance unknown
        out[6] = -1f // tilt unknown
        out[7] = -1f // azimuth unknown
        out[8] = -1f // roll unknown
    }
 }
@@ -1,9 +1,11 @@
 package io.unom.punktfunk
 import androidx.compose.foundation.gestures.awaitEachGesture
-import androidx.compose.foundation.gestures.awaitFirstDown
+import androidx.compose.ui.input.pointer.AwaitPointerEventScope
 import androidx.compose.ui.input.pointer.PointerId
 import androidx.compose.ui.input.pointer.PointerInputChange
 import androidx.compose.ui.input.pointer.PointerInputScope
 import androidx.compose.ui.input.pointer.PointerType
 import androidx.compose.ui.input.pointer.changedToDownIgnoreConsumed
 import androidx.compose.ui.input.pointer.changedToUpIgnoreConsumed
 import androidx.compose.ui.input.pointer.positionChanged
@@ -56,7 +58,26 @@ private const val ACCEL_MAX = 3.0f
 * normalizes and maps into the output). On teardown (stream leaves composition) every still-held
 * contact is lifted so nothing stays stuck on the host.
 */
-internal suspend fun PointerInputScope.streamTouchPassthrough(handle: Long) {
+/** Whether this change belongs to the stylus lane (only when a pen-capable host is live). */
 private fun isStylus(c: PointerInputChange, stylus: StylusStream?): Boolean =
    stylus != null && (c.type == PointerType.Stylus || c.type == PointerType.Eraser)
 /** [awaitFirstDown] with the stylus lane split out: pen events feed [stylus] and never start a
 *  mouse/touch gesture. Toward a pen-less host ([stylus] == null) a stylus stays a finger. */
 private suspend fun AwaitPointerEventScope.awaitFirstFingerDown(
    stylus: StylusStream?,
 ): PointerInputChange {
    while (true) {
        val ev = awaitPointerEvent()
        stylus?.intercept(ev, size)
        val down = ev.changes.firstOrNull {
            it.changedToDownIgnoreConsumed() && !isStylus(it, stylus)
        }
        if (down != null) return down
    }
 }
 internal suspend fun PointerInputScope.streamTouchPassthrough(handle: Long, stylus: StylusStream?) {
    val ids = mutableMapOf<PointerId, Int>()
    fun alloc(p: PointerId): Int {
        var id = 0
@@ -68,10 +89,12 @@ internal suspend fun PointerInputScope.streamTouchPassthrough(handle: Long) {
        awaitPointerEventScope {
            while (true) {
                val ev = awaitPointerEvent()
                stylus?.intercept(ev, size)
                val sw = size.width
                val sh = size.height
                if (sw <= 0 || sh <= 0) continue
                for (c in ev.changes) {
                    if (isStylus(c, stylus)) continue // the pen plane owns it
                    val x = c.position.x.roundToInt().coerceIn(0, sw - 1)
                    val y = c.position.y.roundToInt().coerceIn(0, sh - 1)
                    when {
@@ -98,10 +121,13 @@ internal suspend fun PointerInputScope.streamTouchPassthrough(handle: Long) {
 internal suspend fun PointerInputScope.streamTouchInput(
    handle: Long,
    stylus: StylusStream?,
    trackpad: Boolean,
    invertScroll: Boolean,
    onCycleStats: () -> Unit,
    onKeyboard: (show: Boolean) -> Unit,
 ) {
    val scrollDir = if (invertScroll) -1 else 1
    var lastTapUp = 0L
    var lastTapX = 0f
    var lastTapY = 0f
@@ -118,7 +144,7 @@ internal suspend fun PointerInputScope.streamTouchInput(
        )
    }
    awaitEachGesture {
-        val down = awaitFirstDown(requireUnconsumed = false)
+        val down = awaitFirstFingerDown(stylus)
        val startX = down.position.x
        val startY = down.position.y
        // A touch landing just after a quick tap nearby = tap-and-drag: hold the left
@@ -155,7 +181,8 @@ internal suspend fun PointerInputScope.streamTouchInput(
        while (true) {
            val ev = awaitPointerEvent()
-            val pressed = ev.changes.filter { it.pressed }
+            stylus?.intercept(ev, size)
            val pressed = ev.changes.filter { it.pressed && !isStylus(it, stylus) }
            if (pressed.isEmpty()) {
                upTime = ev.changes.firstOrNull()?.uptimeMillis ?: upTime
                break
@@ -184,12 +211,12 @@ internal suspend fun PointerInputScope.streamTouchInput(
                val sy = ((prevCy - cy) / SCROLL_DIV).toInt() // finger up → wheel up
                val sx = ((cx - prevCx) / SCROLL_DIV).toInt()
                if (sy != 0) {
-                    NativeBridge.nativeSendScroll(handle, 0, sy * 120)
+                    NativeBridge.nativeSendScroll(handle, 0, sy * 120 * scrollDir)
                    prevCy = cy
                    moved = true
                }
                if (sx != 0) {
-                    NativeBridge.nativeSendScroll(handle, 1, sx * 120)
+                    NativeBridge.nativeSendScroll(handle, 1, sx * 120 * scrollDir)
                    prevCx = cx
                    moved = true
                }
@@ -106,6 +106,17 @@ object Keymap {
        KeyEvent.KEYCODE_DPAD_UP -> 0x26
        KeyEvent.KEYCODE_DPAD_RIGHT -> 0x27
        KeyEvent.KEYCODE_DPAD_DOWN -> 0x28
        // TV-remote SELECT = Enter (a gamepad's press routes via SOURCE_GAMEPAD before this).
        KeyEvent.KEYCODE_DPAD_CENTER -> 0x0D
        // Consumer/media keys — forwarded to the host while streaming (volume stays local:
        // MainActivity's pass-through list wins before the map is consulted).
        KeyEvent.KEYCODE_MEDIA_PLAY_PAUSE,
        KeyEvent.KEYCODE_MEDIA_PLAY,
        KeyEvent.KEYCODE_MEDIA_PAUSE -> 0xB3 // VK_MEDIA_PLAY_PAUSE
        KeyEvent.KEYCODE_MEDIA_NEXT -> 0xB0 // VK_MEDIA_NEXT_TRACK
        KeyEvent.KEYCODE_MEDIA_PREVIOUS -> 0xB1 // VK_MEDIA_PREV_TRACK
        KeyEvent.KEYCODE_MEDIA_STOP -> 0xB2 // VK_MEDIA_STOP
        // Modifiers (L/R-specific VKs; the host folds the generic ones onto the left variant)
        KeyEvent.KEYCODE_SHIFT_LEFT -> 0xA0
@@ -287,6 +287,66 @@ object NativeBridge {
    /** One key transition. vk: Windows VK (0 = dropped by Rust). mods: VK modifier mask (0 for now). */
    external fun nativeSendKey(handle: Long, vk: Int, down: Boolean, mods: Int)
    /**
     * Whether the host advertised full-fidelity stylus injection (`HOST_CAP_PEN`) — the gate
     * for splitting stylus pointers out of the touch path onto the pen plane. False on `0`.
     */
    external fun nativeHostSupportsPen(handle: Long): Boolean
    /**
     * One stylus batch of STATE-FULL samples (the pen plane; design/pen-tablet-input.md §7):
     * [count] × 10 floats, oldest first — `[state, tool, x, y, pressure, distance, tilt_deg,
     * azimuth_deg, roll_deg, dt_us]`. `state` = the wire in-range/touching/barrel bits; `tool`
     * 0=pen 1=eraser; x/y/pressure/distance normalized 0..1; distance/tilt/azimuth/roll < 0 =
     * unknown. Send only when [nativeHostSupportsPen]; repeat the last sample ≤100 ms while the
     * pen is in range (the host force-releases a silent stroke after 200 ms).
     */
    external fun nativeSendPen(handle: Long, samples: FloatArray, count: Int)
    /**
     * Whether the host advertised committed-text injection (`HOST_CAP_TEXT_INPUT`) — its inject
     * backend can type Unicode text directly. Picks the real IME `InputConnection` (autocorrect,
     * gesture typing, non-Latin scripts) over the TYPE_NULL raw-key fallback. False on `0`.
     */
    external fun nativeTextInputSupported(handle: Long): Boolean
    /**
     * Committed IME text → one `TextInput` wire event per Unicode scalar, in order. Control
     * characters are skipped natively (Enter/Backspace ride [nativeSendKey]). Only meaningful
     * when [nativeTextInputSupported] returned true — older hosts ignore the events.
     */
    external fun nativeSendText(handle: Long, text: String)
    // ---- Shared clipboard (text v1): Kotlin drives ClipboardManager, Rust the protocol ----
    // Opt-in per session (nativeClipControl). Local copies are announced as lazy offers; bytes
    // cross only when the host pastes (a "fetch:" event answered by nativeClipServeText). Host
    // copies arrive as "offer:" events, fetched eagerly into the system clipboard.
    /** Whether the host advertised a working shared-clipboard service (HOST_CAP_CLIPBOARD). */
    external fun nativeClipSupported(handle: Long): Boolean
    /** Session-level clipboard opt-in/out; nothing happens until enabled=true crosses. */
    external fun nativeClipControl(handle: Long, enabled: Boolean)
    /** Announce "this device's clipboard now holds text". [seq]: monotonic, newest wins. */
    external fun nativeClipOfferText(handle: Long, seq: Int)
    /** Pull the text of the host's offer [seq] → transfer id echoed on "data:"/"error:", or -1. */
    external fun nativeClipFetchText(handle: Long, seq: Int): Int
    /** Answer a "fetch:" event with the clipboard's current text (the host is pasting). */
    external fun nativeClipServeText(handle: Long, reqId: Int, text: String)
    /** Abort a clipboard transfer by id (either direction). */
    external fun nativeClipCancel(handle: Long, id: Int)
    /**
     * Block ≤250 ms for the next clipboard event, as a compact string: `state:<0|1>` ·
     * `offer:<seq>:<hasText>` · `fetch:<reqId>` · `data:<xferId>:<text>` · `cancel:<id>` ·
     * `error:<id>:<code>` · `closed` (session gone) — null on timeout. Dedicated poll thread.
     */
    external fun nativeNextClip(handle: Long): String?
    // ---- Gamepad: each controller forwarded on its own wire pad index (0..15, low byte of flags) ----
    // The pad index is assigned per Android device by GamepadRouter; a single controller lands on 0,
    // so its wire is byte-identical to the old single-pad path. The core folds the per-transition
@@ -0,0 +1,182 @@
 //! Shared-clipboard plane (text-only v1): Kotlin drives the Android `ClipboardManager`, these
 //! shims drive [`punktfunk_core::client::NativeClient`]'s clipboard surface.
 //!
 //! Model (mirrors the desktop clients): opt-in via `nativeClipControl(true)`; local copies are
 //! announced lazily as format-list offers (`nativeClipOfferText`) and the bytes cross only when
 //! the host pastes (a `fetch` event answered by `nativeClipServeText`); a host copy arrives as an
 //! `offer` event, which the Kotlin side fetches eagerly (Android's clipboard has no lazy provider
 //! path worth the complexity) and lands in the system clipboard on the `data` event.
 //!
 //! Events cross to Kotlin as compact strings from the blocking `nativeNextClip` poll (drained on
 //! a dedicated thread, same pattern as `nativeNextRumble`):
 //! `state:<0|1>` · `offer:<seq>:<has_text 0|1>` · `fetch:<req_id>` · `data:<xfer_id>:<text>` ·
 //! `cancel:<id>` · `error:<id>:<code>` · `closed` — null on a poll timeout. Non-text fetch
 //! requests are cancelled natively (only text is ever offered, so they shouldn't occur).
 use std::time::Duration;
 use jni::objects::{JObject, JString};
 use jni::sys::{jboolean, jint, jlong, jstring};
 use jni::JNIEnv;
 use punktfunk_core::clipboard::ClipEventCore;
 use punktfunk_core::error::PunktfunkError;
 use punktfunk_core::quic::{ClipKind, CLIP_FILE_INDEX_NONE, HOST_CAP_CLIPBOARD};
 use super::SessionHandle;
 /// The portable wire MIME both ends map to their platform text type.
 const TEXT_MIME: &str = "text/plain;charset=utf-8";
 /// Deref the opaque handle (`0` → `None`).
 ///
 /// SAFETY: live handle per the nativeConnect/nativeClose contract; every method used is `&self`
 /// on the `Sync` connector.
 fn client(handle: jlong) -> Option<&'static SessionHandle> {
    if handle == 0 {
        return None;
    }
    // SAFETY: see the function docs — the Kotlin side guarantees the handle outlives the call.
    Some(unsafe { &*(handle as *const SessionHandle) })
 }
 /// `NativeBridge.nativeClipSupported(handle)` — the host advertised `HOST_CAP_CLIPBOARD`.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeClipSupported(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
 ) -> jboolean {
    client(handle).map_or(0, |h| {
        u8::from(h.client.host_caps() & HOST_CAP_CLIPBOARD != 0)
    })
 }
 /// `NativeBridge.nativeClipControl(handle, enabled)` — session-level opt-in/out. Nothing
 /// clipboard-related happens on either side until an `enabled: true` crosses.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeClipControl(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
    enabled: jboolean,
 ) {
    if let Some(h) = client(handle) {
        let _ = h.client.clip_control(enabled != 0, 0);
    }
 }
 /// `NativeBridge.nativeClipOfferText(handle, seq)` — announce "the Android clipboard now holds
 /// text" (format list only; bytes cross when the host fetches). `seq` is Kotlin's monotonic
 /// counter, newest wins.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeClipOfferText(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
    seq: jint,
 ) {
    if let Some(h) = client(handle) {
        let _ = h.client.clip_offer(
            seq as u32,
            vec![ClipKind {
                mime: TEXT_MIME.into(),
                size_hint: 0,
            }],
        );
    }
 }
 /// `NativeBridge.nativeClipFetchText(handle, seq)` — pull the text of the host's offer `seq`.
 /// Returns the transfer id echoed on the matching `data:`/`error:` event, or −1.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeClipFetchText(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
    seq: jint,
 ) -> jint {
    client(handle)
        .and_then(|h| {
            h.client
                .clip_fetch(seq as u32, TEXT_MIME.into(), CLIP_FILE_INDEX_NONE)
                .ok()
        })
        .map_or(-1, |xfer| xfer as jint)
 }
 /// `NativeBridge.nativeClipServeText(handle, reqId, text)` — answer a `fetch:` event with the
 /// clipboard's current text (the host is pasting our offer).
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeClipServeText(
    mut env: JNIEnv,
    _this: JObject,
    handle: jlong,
    req_id: jint,
    text: JString,
 ) {
    let Some(h) = client(handle) else { return };
    let Ok(s) = env.get_string(&text) else {
        let _ = h.client.clip_cancel(req_id as u32);
        return;
    };
    let _ = h
        .client
        .clip_serve(req_id as u32, String::from(s).into_bytes(), true);
 }
 /// `NativeBridge.nativeClipCancel(handle, id)` — abort a transfer (either direction).
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeClipCancel(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
    id: jint,
 ) {
    if let Some(h) = client(handle) {
        let _ = h.client.clip_cancel(id as u32);
    }
 }
 /// `NativeBridge.nativeNextClip(handle)` — block ≤250 ms for the next clipboard event, encoded
 /// as a compact string (module docs); null on timeout, `"closed"` once the session is gone.
 /// Call from a dedicated poll thread.
 ///
 /// Text payloads ride `data:<xfer_id>:<text>` decoded lossily — safe because the phase-0
 /// clipboard task delivers a whole payload in ONE event (`last = true`), so a chunk boundary
 /// can never split a UTF-8 sequence.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeNextClip(
    env: JNIEnv,
    _this: JObject,
    handle: jlong,
 ) -> jstring {
    let Some(h) = client(handle) else {
        return std::ptr::null_mut();
    };
    let msg = match h.client.next_clip(Duration::from_millis(250)) {
        Ok(ClipEventCore::State { enabled, .. }) => format!("state:{}", u8::from(enabled)),
        Ok(ClipEventCore::RemoteOffer { seq, kinds }) => {
            let has_text = kinds.iter().any(|k| k.mime.starts_with("text/plain"));
            format!("offer:{seq}:{}", u8::from(has_text))
        }
        Ok(ClipEventCore::FetchRequest { req_id, mime, .. }) => {
            if mime.starts_with("text/plain") {
                format!("fetch:{req_id}")
            } else {
                // We only ever offer text; cancel anything else rather than stall the host.
                let _ = h.client.clip_cancel(req_id);
                return std::ptr::null_mut();
            }
        }
        Ok(ClipEventCore::Data { xfer_id, bytes, .. }) => {
            format!("data:{xfer_id}:{}", String::from_utf8_lossy(&bytes))
        }
        Ok(ClipEventCore::Cancelled { id }) => format!("cancel:{id}"),
        Ok(ClipEventCore::Error { id, code }) => format!("error:{id}:{code}"),
        Err(PunktfunkError::NoFrame) => return std::ptr::null_mut(),
        Err(_) => "closed".into(),
    };
    env.new_string(msg)
        .map(|s| s.into_raw())
        .unwrap_or(std::ptr::null_mut())
 }
@@ -201,6 +201,9 @@ pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeConnect<'lo
        // No display-volume forwarding from Android yet (the panel tone-maps PQ itself via the
        // Surface dataspace + static metadata) — the host keeps its virtual-display EDID defaults.
        None,
        // No non-video caps: this client does not render the host cursor locally (no shape/state
        // planes in the jni surface), so advertising CLIENT_CAP_CURSOR would stream cursor-less.
        0,
        launch,   // a store-qualified library id to boot into a game, or None for the desktop
        pin,      // Some → Crypto on host-fp mismatch
        identity, // owned (cert, key) PEM, or None (anonymous)
@@ -6,11 +6,14 @@
 //! conventions: buttons 1=left/2=middle/3=right/4=X1/5=X2; scroll axis 0=vertical/1=horizontal,
 //! signed 120-unit delta, +=up/right; keys are Windows VK (mapped from KEYCODE_* on the Kotlin side).
-use jni::objects::{JByteBuffer, JObject};
+use jni::objects::{JByteBuffer, JFloatArray, JObject, JString};
 use jni::sys::{jboolean, jint, jlong};
 use jni::JNIEnv;
 use punktfunk_core::input::{InputEvent, InputKind};
-use punktfunk_core::quic::{RichInput, HID_REPORT_MAX};
+use punktfunk_core::quic::{
    PenSample, PenTool, RichInput, HID_REPORT_MAX, HOST_CAP_PEN, HOST_CAP_TEXT_INPUT,
    PEN_ANGLE_UNKNOWN, PEN_BATCH_MAX, PEN_DISTANCE_UNKNOWN, PEN_TILT_UNKNOWN,
 };
 use super::SessionHandle;
@@ -145,6 +148,132 @@ pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeSendKey(
    send_event(handle, kind, vk as u32, 0, 0, mods as u32);
 }
 /// `NativeBridge.nativeTextInputSupported(handle)` — whether the host advertised
 /// `HOST_CAP_TEXT_INPUT` (its inject backend types committed text), so the Kotlin side can pick
 /// the real IME `InputConnection` over the TYPE_NULL raw-key fallback. `0` handle → false.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeTextInputSupported(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
 ) -> jboolean {
    if handle == 0 {
        return 0;
    }
    // SAFETY: live handle per the nativeConnect/nativeClose contract; host_caps is &self.
    let h = unsafe { &*(handle as *const SessionHandle) };
    u8::from(h.client.host_caps() & HOST_CAP_TEXT_INPUT != 0)
 }
 /// `NativeBridge.nativeHostSupportsPen(handle)` — the host advertised `HOST_CAP_PEN`, so the
 /// Kotlin side splits stylus pointers out of the touch path onto the pen plane
 /// (design/pen-tablet-input.md §7). `0` handle → false.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeHostSupportsPen(
    _env: JNIEnv,
    _this: JObject,
    handle: jlong,
 ) -> jboolean {
    if handle == 0 {
        return 0;
    }
    // SAFETY: live handle per the nativeConnect/nativeClose contract; host_caps is &self.
    let h = unsafe { &*(handle as *const SessionHandle) };
    u8::from(h.client.host_caps() & HOST_CAP_PEN != 0)
 }
 /// Floats per sample in the `nativeSendPen` flat array.
 const PEN_JNI_STRIDE: usize = 10;
 /// `NativeBridge.nativeSendPen(handle, samples, count)` — one stylus batch of STATE-FULL
 /// samples, `count` × [`PEN_JNI_STRIDE`] floats, oldest first:
 /// `[state, tool, x, y, pressure, distance, tilt_deg, azimuth_deg, roll_deg, dt_us]`.
 /// `state` = the wire `PEN_*` bits; `tool` 0=pen 1=eraser; `x`/`y`/`pressure`/`distance`
 /// normalized 0..1; `distance`/`tilt_deg`/`azimuth_deg`/`roll_deg` < 0 = unknown. Call only
 /// against a [`nativeHostSupportsPen`] host; the client heartbeats the last sample ≤100 ms
 /// while in range (Kotlin side — see `StylusStream`).
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeSendPen(
    env: JNIEnv,
    _this: JObject,
    handle: jlong,
    samples: JFloatArray,
    count: jint,
 ) {
    if handle == 0 || count <= 0 {
        return;
    }
    let count = (count as usize).min(PEN_BATCH_MAX);
    let mut buf = [0f32; PEN_BATCH_MAX * PEN_JNI_STRIDE];
    let flat = &mut buf[..count * PEN_JNI_STRIDE];
    if env.get_float_array_region(&samples, 0, flat).is_err() {
        return; // short array — a bridge bug, never worth a crash on the input path
    }
    let mut batch = [PenSample::default(); PEN_BATCH_MAX];
    for (slot, s) in batch.iter_mut().zip(flat.chunks_exact(PEN_JNI_STRIDE)) {
        if !s[2].is_finite() || !s[3].is_finite() {
            return; // never forward a NaN coordinate
        }
        *slot = PenSample {
            state: s[0] as u8,
            tool: if s[1] as u8 == 1 {
                PenTool::Eraser
            } else {
                PenTool::Pen
            },
            x: s[2].clamp(0.0, 1.0),
            y: s[3].clamp(0.0, 1.0),
            pressure: (s[4].clamp(0.0, 1.0) * 65535.0) as u16,
            distance: if s[5] < 0.0 {
                PEN_DISTANCE_UNKNOWN
            } else {
                (s[5].clamp(0.0, 1.0) * 65534.0) as u16
            },
            tilt_deg: if s[6] < 0.0 {
                PEN_TILT_UNKNOWN
            } else {
                (s[6].clamp(0.0, 90.0)) as u8
            },
            azimuth_deg: if s[7] < 0.0 {
                PEN_ANGLE_UNKNOWN
            } else {
                (s[7] as u16) % 360
            },
            roll_deg: if s[8] < 0.0 {
                PEN_ANGLE_UNKNOWN
            } else {
                (s[8] as u16) % 360
            },
            dt_us: s[9].clamp(0.0, 65535.0) as u16,
        };
    }
    // SAFETY: live handle per the nativeConnect/nativeClose contract; send_pen is &self.
    let h = unsafe { &*(handle as *const SessionHandle) };
    let _ = h.client.send_pen(&batch[..count]);
 }
 /// `NativeBridge.nativeSendText(handle, text)` — committed IME text, one `TextInput` event per
 /// Unicode scalar (`code` = the scalar; multi-char commits are consecutive events in order).
 /// Control characters are skipped — Enter/Backspace/Tab ride the VK key path. Call only when
 /// [`Java_io_unom_punktfunk_kit_NativeBridge_nativeTextInputSupported`] returned true.
 #[no_mangle]
 pub extern "system" fn Java_io_unom_punktfunk_kit_NativeBridge_nativeSendText(
    mut env: JNIEnv,
    _this: JObject,
    handle: jlong,
    text: JString,
 ) {
    if handle == 0 {
        return;
    }
    let Ok(s) = env.get_string(&text) else {
        return;
    };
    for ch in String::from(s).chars().filter(|c| !c.is_control()) {
        send_event(handle, InputKind::TextInput, ch as u32, 0, 0, 0);
    }
 }
 // ---- Gamepad: Kotlin captures (KeyEvent/MotionEvent) → NativeClient::send_input ---------------
 // Multi-pad model: each physical controller is forwarded on its own wire pad index (0..15), carried
 // in the low byte of `flags` on every per-pad event — the Kotlin side (`GamepadRouter`) assigns a
@@ -17,6 +17,7 @@
 //! TODO(M4 Android stage 1): client→host DualSense rich input (`send_rich_input`), mode
 //! renegotiation. Port the remaining orchestration from `clients/linux`.
 mod clipboard;
 mod connect;
 mod input;
 mod planes;
@@ -302,13 +302,26 @@ final class SessionModel: ObservableObject {
            if preferredCodec == PunktfunkConnection.codecPyroWave, MetalWaveletDecoder.supported {
                videoCodecs |= PunktfunkConnection.codecPyroWave
            }
            // Cursor channel (remote-desktop-sweep M2, macOS): sessions STARTING in the desktop
            // mouse model advertise local cursor rendering — the host then stops compositing
            // the pointer and forwards shape/state, which StreamView draws as the real
            // NSCursor. Capture-mode sessions keep today's composited pointer.
            #if os(macOS)
            let clientCaps: UInt8 =
                (MouseInputMode(
                    rawValue: UserDefaults.standard.string(forKey: DefaultsKey.mouseMode) ?? "")
                    ?? .capture) == .desktop ? 0x01 : 0
            #else
            let clientCaps: UInt8 = 0
            #endif
            let result = Result { try PunktfunkConnection(
                host: host.address, port: host.port,
                width: width, height: height, refreshHz: hz,
                pinSHA256: pin, identity: identity, compositor: compositor,
                gamepad: gamepad, bitrateKbps: bitrateKbps, videoCaps: videoCaps,
                audioChannels: audioChannels,
-                videoCodecs: videoCodecs, preferredCodec: preferredCodec, launchID: launchID,
+                videoCodecs: videoCodecs, preferredCodec: preferredCodec,
                clientCaps: clientCaps, launchID: launchID,
                // Delegated approval: the host holds this connect open until the operator approves
                // it (~180 s) — outwait that window so a slow approval still lands here. Normal
                // connects keep the snappy default.
@@ -221,6 +221,9 @@ public final class PunktfunkConnection {
    /// core). The clip *sends* (`clipControl`/`clipOffer`/`clipServe`…) share this lock too:
    /// they're quick non-blocking enqueues, and a single lock keeps close() ordering simple.
    private let clipboardLock = NSLock()
    /// Serializes the (single) cursor pull thread against close() — both cursor planes are
    /// drained by ONE thread, so one lock covers them.
    private let cursorLock = NSLock()
    /// Negotiated session mode (host-confirmed).
    public private(set) var width: UInt32 = 0
@@ -381,6 +384,106 @@ public final class PunktfunkConnection {
    public var hostSupportsClipboard: Bool {
        hostCaps & UInt8(PUNKTFUNK_HOST_CAP_CLIPBOARD) != 0
    }
    /// The host answered `HOST_CAP_CURSOR`: it stopped compositing the pointer and forwards
    /// shape/state on the cursor planes — the client MUST draw the cursor locally.
    /// `0x08` — the bit moved when `HOST_CAP_TEXT_INPUT` claimed `0x04` on main; testing the
    /// old bit would mistake a text-input-capable host (e.g. Windows) for a cursor grant.
    public var hostSupportsCursor: Bool {
        hostCaps & 0x08 != 0
    }
    /// The host injects full-fidelity stylus input (`HOST_CAP_PEN`) — the gate for splitting
    /// Apple Pencil out of the touch path onto the pen plane (``sendPen(_:)``).
    public var hostSupportsPen: Bool {
        hostCaps & UInt8(PUNKTFUNK_HOST_CAP_PEN) != 0
    }
    /// One forwarded host-cursor shape (the cursor channel, ABI v11): straight-alpha RGBA,
    /// `rgba.count == width * height * 4`, hotspot within the bitmap. Cache by `serial` —
    /// states reference shapes by it and a re-shown serial never resends pixels.
    public struct CursorShapeEvent: Sendable {
        public let serial: UInt32
        public let width: Int
        public let height: Int
        public let hotX: Int
        public let hotY: Int
        public let rgba: Data
    }
    /// Per-host-tick cursor state: position (host video px, the pointer/hotspot point),
    /// visibility, and the host-driven relative-mode hint (an app grabbed/hid the pointer ⇒
    /// run captured relative; clear ⇒ absolute, reappearing at `x`/`y`). Latest-wins.
    public struct CursorStateEvent: Sendable {
        public let serial: UInt32
        public let visible: Bool
        public let relativeHint: Bool
        public let x: Int32
        public let y: Int32
    }
    /// Pull the next forwarded cursor SHAPE (nil = timeout). Only a session connected with
    /// `clientCaps` cursor bit against a `hostSupportsCursor` host receives any. Drain shape
    /// AND state from ONE dedicated cursor thread (they share a lock).
    public func nextCursorShape(timeoutMs: UInt32 = 0) throws -> CursorShapeEvent? {
        cursorLock.lock()
        defer { cursorLock.unlock() }
        guard let h = liveHandle() else { throw PunktfunkClientError.closed }
        var out = PunktfunkCursorShape()
        let rc = punktfunk_connection_next_cursor_shape(h, &out, timeoutMs)
        switch rc {
        case statusOK:
            // Copy out of the ABI borrow (valid until the next shape call) immediately.
            let bytes = out.rgba.map { Data(bytes: $0, count: Int(out.len)) } ?? Data()
            return CursorShapeEvent(
                serial: out.serial, width: Int(out.w), height: Int(out.h),
                hotX: Int(out.hot_x), hotY: Int(out.hot_y), rgba: bytes)
        case statusNoFrame:
            return nil
        case statusClosed:
            throw PunktfunkClientError.closed
        default:
            throw PunktfunkClientError.status(rc)
        }
    }
    /// Pull the next cursor STATE (nil = timeout). Latest-wins — drain the queue and apply
    /// only the newest. Same thread + gate as [`nextCursorShape`].
    public func nextCursorState(timeoutMs: UInt32 = 0) throws -> CursorStateEvent? {
        cursorLock.lock()
        defer { cursorLock.unlock() }
        guard let h = liveHandle() else { throw PunktfunkClientError.closed }
        var out = PunktfunkCursorState()
        let rc = punktfunk_connection_next_cursor_state(h, &out, timeoutMs)
        switch rc {
        case statusOK:
            return CursorStateEvent(
                serial: out.serial,
                visible: out.flags & 0x01 != 0,
                relativeHint: out.flags & 0x02 != 0,
                x: out.x, y: out.y)
        case statusNoFrame:
            return nil
        case statusClosed:
            throw PunktfunkClientError.closed
        default:
            throw PunktfunkClientError.status(rc)
        }
    }
    /// Tell the host who renders the pointer (the §8 mid-stream mouse-model flip, ABI v12):
    /// `clientDraws = true` — this client draws it locally (the desktop mouse model; the host
    /// excludes the pointer from the video and forwards shape/state); `false` — the host
    /// composites it into the video (the capture model, full fidelity). Idempotent,
    /// latest-wins; harmless against hosts without the cursor cap. Fire-and-forget — errors
    /// are swallowed (a closed session is the only failure and it moots the flip).
    public func setCursorRender(clientDraws: Bool) {
        cursorLock.lock()
        defer { cursorLock.unlock() }
        guard let h = liveHandle() else { return }
        _ = punktfunk_connection_set_cursor_render(h, clientDraws)
    }
    /// The resolved codec as a `VideoCodec` (H.264 / HEVC / AV1) — drives the bitstream framing
    /// (Annex-B NAL parsing vs the AV1 OBU repack).
    public var videoCodec: VideoCodec { VideoCodec(wire: resolvedCodec) }
@@ -417,6 +520,7 @@ public final class PunktfunkConnection {
        audioChannels: UInt8 = 2,
        videoCodecs: UInt8 = 0x02, // PUNKTFUNK_CODEC_HEVC — the codecs this client can decode
        preferredCodec: UInt8 = 0, // 0 = auto; else PUNKTFUNK_CODEC_* soft preference
        clientCaps: UInt8 = 0, // ABI v11: PUNKTFUNK_CLIENT_CAP_CURSOR = render the host cursor locally
        launchID: String? = nil,
        timeoutMs: UInt32 = 10_000
    ) throws {
@@ -436,18 +540,18 @@ public final class PunktfunkConnection {
                    withOptionalCString(launchID) { launch in
                        if let pin = pinSHA256 {
                            return pin.withUnsafeBytes { p in
-                                punktfunk_connect_ex8(
+                                punktfunk_connect_ex9(
                                    cs, port, width, height, refreshHz, compositor.rawValue,
                                    gamepad.rawValue, bitrateKbps, videoCaps, audioChannels,
-                                    videoCodecs, preferredCodec, launch,
+                                    videoCodecs, preferredCodec, clientCaps, launch,
                                    p.bindMemory(to: UInt8.self).baseAddress, &observed,
                                    cert, key, timeoutMs, &connectStatus)
                            }
                        }
-                        return punktfunk_connect_ex8(
+                        return punktfunk_connect_ex9(
                            cs, port, width, height, refreshHz, compositor.rawValue,
                            gamepad.rawValue, bitrateKbps, videoCaps, audioChannels,
-                            videoCodecs, preferredCodec, launch,
+                            videoCodecs, preferredCodec, clientCaps, launch,
                            nil, &observed, cert, key, timeoutMs, &connectStatus)
                    }
                }
@@ -1037,6 +1141,19 @@ public final class PunktfunkConnection {
        _ = punktfunk_connection_send_input(h, &ev)
    }
    /// Send one stylus sample batch (≤ `PUNKTFUNK_PEN_BATCH_MAX`, oldest first) on the pen
    /// plane. Gate on ``hostSupportsPen`` — the core refuses toward a host without the cap.
    /// Thread-safe; silently dropped after close (input is lossy by design).
    public func sendPen(_ samples: [PunktfunkPenSample]) {
        guard !samples.isEmpty else { return }
        abiLock.lock()
        defer { abiLock.unlock() }
        guard let h = handle, !closeRequested else { return }
        samples.withUnsafeBufferPointer { buf in
            _ = punktfunk_connection_send_pen(h, buf.baseAddress, UInt32(buf.count))
        }
    }
    /// Signal a **deliberate** user-initiated quit before ``close()``: the connection closes with
    /// `QUIT_CLOSE_CODE` (81) so the host tears the session down immediately instead of holding the
    /// keep-alive linger for a reconnect. Call only from an explicit "Disconnect" action — NOT from a
@@ -1059,10 +1176,12 @@ public final class PunktfunkConnection {
        feedbackLock.lock()
        statsLock.lock()
        clipboardLock.lock()
        cursorLock.lock()
        abiLock.lock()
        let h = handle
        handle = nil
        abiLock.unlock()
        cursorLock.unlock()
        clipboardLock.unlock()
        statsLock.unlock()
        feedbackLock.unlock()
@@ -0,0 +1,268 @@
 // Apple Pencil → state-full wire pen samples (design/pen-tablet-input.md §7).
 //
 // Every sample carries the COMPLETE pen state (in-range/touching/buttons + all axes) — the
 // host diffs consecutive samples and synthesizes down/up/button transitions itself, so a lost
 // datagram self-heals and this file never sends edge events. Three sources feed one stream:
 // UITouch contacts (with coalesced samples for full 240 Hz fidelity), the hover gesture
 // (zOffset > 0 distinguishes a hovering Pencil from a trackpad pointer), and
 // UIPencilInteraction (squeeze held → barrel 1, double-tap → a momentary barrel 2 —
 // Apple Pencil has no hardware eraser end or barrel buttons, so these mappings are how
 // host-side apps get their stylus button/eraser affordances).
 //
 // HEARTBEAT (wire contract — see `PunktfunkPenSample` in punktfunk_core.h): while the pen is
 // in range or touching, the last sample repeats every ≤100 ms even when nothing changed.
 // UIKit is silent for a stationary Pencil, and the host force-releases the stroke after
 // 200 ms without samples (its dead-client failsafe) — the timer keeps a held stroke alive.
 #if os(iOS)
 import PunktfunkCore
 import UIKit
 final class PencilStream: NSObject, UIPencilInteractionDelegate {
    enum Phase { case down, move, up, cancel }
    /// One assembled batch (≤ `PUNKTFUNK_PEN_BATCH_MAX` samples) ready for the connection.
    var send: (([PunktfunkPenSample]) -> Void)?
    /// View-space point → normalized [0,1] video coordinates (the letterbox mapping the
    /// touch path already uses). nil until a mode is negotiated — samples are dropped then.
    var videoNorm: ((CGPoint) -> (Float, Float)?)?
    private var inRange = false
    private var touching = false
    /// Squeeze held (mapped to wire BARREL1).
    private var squeezeHeld = false
    /// Whether this device/Pencil pair has demonstrated hover — decides what a lift means:
    /// hover-capable hardware keeps proximity (the hover recognizer owns the exit), anything
    /// else leaves range on lift so the host never parks a phantom hovering pen.
    private var sawHover = false
    /// A hover gesture is live right now (routes ended-state hover callbacks to us even when
    /// the recognizer's final zOffset reads 0).
    private(set) var hoverActive = false
    private var last = PencilStream.idleSample()
    private var heartbeat: Timer?
    // MARK: - Contact path (UITouch, `.pencil` only)
    func touches(_ touches: Set<UITouch>, event: UIEvent?, phase: Phase, in view: UIView) {
        // At most one Pencil exists; a set with several is UIKit batching phases of the same
        // stylus — the last one carries the freshest state.
        guard let touch = touches.max(by: { $0.timestamp < $1.timestamp }) else { return }
        switch phase {
        case .down, .move:
            touching = true
            inRange = true
            // Coalesced samples restore the Pencil's full capture rate (UIKit delivers at
            // display cadence); oldest first, `dt_us` preserving their spacing.
            let raw = event?.coalescedTouches(for: touch) ?? [touch]
            var batch: [PunktfunkPenSample] = []
            var prevTs: TimeInterval?
            for t in raw.suffix(Int(PUNKTFUNK_PEN_BATCH_MAX)) {
                guard let s = contactSample(t, in: view, prevTs: prevTs) else { continue }
                prevTs = t.timestamp
                batch.append(s)
            }
            emit(batch)
        case .up:
            touching = false
            // Hover-capable hardware: lift back to hover, the recognizer exits range later.
            // Otherwise a lift IS the range exit (mirror of the host's GameStream heuristic).
            inRange = sawHover
            var s = last
            s.pressure = 0
            s.state = stateBits()
            if let posSample = contactSample(touch, in: view, prevTs: nil) {
                s.x = posSample.x
                s.y = posSample.y
            }
            emit([s])
        case .cancel:
            release()
        }
    }
    // MARK: - Hover path (forwarded from the view's hover recognizer)
    /// Returns whether the event was consumed as Pencil hover; `false` hands it back to the
    /// pointer path. A hovering Pencil reports `zOffset > 0`; trackpad/mouse hover is 0.
    func maybeHover(_ r: UIHoverGestureRecognizer, in view: UIView) -> Bool {
        switch r.state {
        case .began, .changed:
            guard r.zOffset > 0 || hoverActive else { return false }
            hoverActive = true
            sawHover = true
            inRange = true
            touching = false
            guard let (x, y) = videoNorm?(r.location(in: view)) else { return true }
            var s = PencilStream.idleSample()
            s.state = stateBits()
            s.x = x
            s.y = y
            s.distance = UInt16((r.zOffset.clamped(to: 0...1) * 65534).rounded())
            s.tilt_deg = Self.tiltDeg(altitude: r.altitudeAngle)
            s.azimuth_deg = Self.azimuthDeg(r.azimuthAngle(in: view))
            if #available(iOS 17.5, *) { s.roll_deg = Self.rollDeg(r.rollAngle) }
            emit([s])
            return true
        case .ended, .cancelled, .failed:
            guard hoverActive else { return false }
            hoverActive = false
            if !touching { release() }
            return true
        default:
            return hoverActive
        }
    }
    // MARK: - UIPencilInteractionDelegate (squeeze → barrel 1 held, tap → barrel 2 click)
    @available(iOS 17.5, *)
    func pencilInteraction(
        _ interaction: UIPencilInteraction, didReceiveSqueeze squeeze: UIPencilInteraction.Squeeze
    ) {
        switch squeeze.phase {
        case .began:
            squeezeHeld = true
        case .ended, .cancelled:
            squeezeHeld = false
        default:
            return
        }
        guard inRange || touching else { return }
        var s = last
        s.state = stateBits()
        emit([s])
    }
    func pencilInteractionDidTap(_ interaction: UIPencilInteraction) {
        guard inRange || touching else { return }
        // A momentary barrel-2 click: press + release as two state-full samples in ONE batch
        // — the host's tracker emits the button press and release in order.
        var press = last
        press.state = stateBits() | UInt8(PUNKTFUNK_PEN_BARREL2)
        var releaseS = last
        releaseS.state = stateBits()
        emit([press, releaseS])
    }
    // MARK: - Lifecycle
    /// Session stop / view teardown: leave range so the host lifts anything held.
    func reset() {
        if inRange || touching { release() }
        sawHover = false
        hoverActive = false
        squeezeHeld = false
    }
    private func release() {
        touching = false
        inRange = false
        var s = last
        s.pressure = 0
        s.state = 0
        emit([s])
    }
    // MARK: - Sample assembly
    private func contactSample(
        _ t: UITouch, in view: UIView, prevTs: TimeInterval?
    ) -> PunktfunkPenSample? {
        guard let (x, y) = videoNorm?(t.location(in: view)) else { return nil }
        var s = PencilStream.idleSample()
        s.state = stateBits()
        s.x = x
        s.y = y
        // maximumPossibleForce is 0 until the system knows the stylus — full force then
        // (binary-stylus semantics, matching the host's unknown-pressure rule).
        let maxForce = t.maximumPossibleForce
        s.pressure =
            maxForce > 0
            ? UInt16((Double(t.force / maxForce).clamped(to: 0...1) * 65535).rounded())
            : UInt16.max
        s.distance = 0
        s.tilt_deg = Self.tiltDeg(altitude: t.altitudeAngle)
        s.azimuth_deg = Self.azimuthDeg(t.azimuthAngle(in: view))
        if #available(iOS 17.5, *) { s.roll_deg = Self.rollDeg(t.rollAngle) }
        if let prevTs {
            s.dt_us = UInt16(((t.timestamp - prevTs) * 1_000_000).clamped(to: 0...65535))
        }
        return s
    }
    private func stateBits() -> UInt8 {
        var bits: UInt8 = 0
        if inRange || touching { bits |= UInt8(PUNKTFUNK_PEN_IN_RANGE) }
        if touching { bits |= UInt8(PUNKTFUNK_PEN_TOUCHING) }
        if squeezeHeld { bits |= UInt8(PUNKTFUNK_PEN_BARREL1) }
        return bits
    }
    private func emit(_ batch: [PunktfunkPenSample]) {
        guard !batch.isEmpty else { return }
        last = batch[batch.count - 1]
        last.dt_us = 0
        send?(batch)
        armHeartbeat()
    }
    /// The ≤100 ms keepalive while in range (see the file header). 80 ms leaves headroom
    /// under the host's 200 ms failsafe even with one lost datagram.
    private func armHeartbeat() {
        heartbeat?.invalidate()
        guard inRange || touching else {
            heartbeat = nil
            return
        }
        heartbeat = Timer.scheduledTimer(withTimeInterval: 0.08, repeats: true) {
            [weak self] _ in
            guard let self, self.inRange || self.touching else {
                self?.heartbeat?.invalidate()
                self?.heartbeat = nil
                return
            }
            self.send?([self.last])
        }
    }
    // MARK: - Angle conversions
    /// Altitude (π/2 = perpendicular) → wire tilt-from-normal in degrees, 0...90.
    private static func tiltDeg(altitude: CGFloat) -> UInt8 {
        UInt8((90 - altitude * 180 / .pi).rounded().clamped(to: 0...90))
    }
    /// Apple azimuth (0 along the view's +x axis, clockwise, y-down) → wire azimuth
    /// (0 = north/up on screen, clockwise): +90° offset.
    private static func azimuthDeg(_ apple: CGFloat) -> UInt16 {
        let deg = (apple * 180 / .pi + 90).truncatingRemainder(dividingBy: 360)
        return UInt16((deg + 360).truncatingRemainder(dividingBy: 360).rounded()) % 360
    }
    /// Pencil Pro roll (radians, −π...π) → wire barrel roll 0...359°.
    private static func rollDeg(_ roll: CGFloat) -> UInt16 {
        let deg = (roll * 180 / .pi).truncatingRemainder(dividingBy: 360)
        return UInt16(((deg + 360).truncatingRemainder(dividingBy: 360)).rounded()) % 360
    }
    /// All-unknown baseline: sentinel angles/distance, tool = pen (the eraser is host-side
    /// state driven by the squeeze/tap mappings, not a hardware end).
    private static func idleSample() -> PunktfunkPenSample {
        PunktfunkPenSample(
            x: 0, y: 0, pressure: 0,
            distance: UInt16(PUNKTFUNK_PEN_DISTANCE_UNKNOWN),
            azimuth_deg: UInt16(PUNKTFUNK_PEN_ANGLE_UNKNOWN),
            roll_deg: UInt16(PUNKTFUNK_PEN_ANGLE_UNKNOWN),
            dt_us: 0, state: 0,
            tool: UInt8(PUNKTFUNK_PEN_TOOL_PEN),
            tilt_deg: UInt8(PUNKTFUNK_PEN_TILT_UNKNOWN),
            _reserved: (0, 0, 0))
    }
 }
 extension Comparable {
    fileprivate func clamped(to range: ClosedRange<Self>) -> Self {
        min(max(self, range.lowerBound), range.upperBound)
    }
 }
 #endif
@@ -219,6 +219,32 @@ public final class StreamLayerView: NSView {
    /// flipped live by ⌃⌥⇧M. A live flip re-engages capture in the new model so
    /// disassociation + the abs/rel choice swap atomically. Main-thread only.
    private var desktopMouse = false
    /// Cursor channel (M2): the host forwards shape/state and WE draw the pointer. Active
    /// when the Welcome carried `HOST_CAP_CURSOR` (only sessions that advertised the client
    /// cap get it). Shapes cache by serial; state is latest-wins. Main-thread only.
    private var cursorChannelActive = false
    /// A forwarded host cursor shape, cached RAW (not as a finished `NSCursor`) so the pointer can be
    /// (re)built at the CURRENT video-fit scale — see `scaledCursor`. The host forwards the bitmap in
    /// host FRAMEBUFFER pixels, whose size tracks the host's display scaling (32 px at 100%, 96 px at
    /// 300% DPI); scaling by the video fit keeps the pointer sized to the streamed desktop at any host
    /// scaling instead of ballooning on a high-DPI host.
    private struct HostCursorShape {
        let cg: CGImage
        let width: Int
        let height: Int
        let hotX: Int
        let hotY: Int
    }
    private var hostCursors: [UInt32: HostCursorShape] = [:]
    private var cursorState: PunktfunkConnection.CursorStateEvent?
    /// Last `CursorRenderMode.clientDraws` told to the host (the §8 mid-stream render flip);
    /// nil = nothing sent yet. Edge-detected by [`reconcileCursorRender`] from the live mouse
    /// model, so the chord, engage/release, and session start all reconcile through one path.
    private var sentClientDraws: Bool?
    /// M3 hint tracking: edge-triggered so a manual ⌃⌥⇧M isn't fought — the override latch
    /// holds until the HOST's intent next changes.
    private var lastHint: Bool?
    private var hintOverride = false
    /// One-shot auto-engage request (stream start, trust confirmed) — attempted as soon
    /// as the view is in a window with real bounds, then dropped, so it can never fire
    /// surprisingly later (e.g. on a resize).
@@ -456,6 +482,7 @@ public final class StreamLayerView: NSView {
        window?.makeFirstResponder(self)
        window?.invalidateCursorRects(for: self) // desktop model: hide-over-view engages
        notifyCaptureChange(true)
        reconcileCursorRender()
    }
    private func releaseCapture() {
@@ -466,6 +493,7 @@ public final class StreamLayerView: NSView {
        captured = false
        window?.invalidateCursorRects(for: self)
        notifyCaptureChange(false)
        reconcileCursorRender() // released ⇒ the host composites the pointer again
    }
    /// A fully transparent cursor for the desktop mouse model's hide-over-view rect —
@@ -480,12 +508,179 @@ public final class StreamLayerView: NSView {
    /// globally via `CursorCapture` (the pointer can't leave the view there).
    override public func resetCursorRects() {
        if captured && desktopMouse {
            // Cursor channel active: wear the HOST's pointer shape (it is no longer in the
            // video); hidden host pointer (or no shape yet) = invisible. Without the channel,
            // M1 behavior: invisible local cursor, the composited host cursor is the visible one.
            if cursorChannelActive, let st = cursorState, st.visible,
               let shape = hostCursors[st.serial] {
                addCursorRect(bounds, cursor: scaledCursor(shape))
            } else {
                addCursorRect(bounds, cursor: Self.invisibleCursor)
            }
        } else {
            super.resetCursorRects()
        }
    }
    /// Tell the host who renders the pointer (the §8 mid-stream render flip): we draw it only
    /// while the DESKTOP model is engaged (the local OS cursor wears the host shape); under
    /// the capture model — and while released — the host composites it into the video (full
    /// fidelity, the pre-channel look). One edge-detected reconciler, called from every
    /// transition (chord, engage/release, session start).
    private func reconcileCursorRender() {
        guard cursorChannelActive, let connection else { return }
        let clientDraws = captured && desktopMouse
        guard sentClientDraws != clientDraws else { return }
        sentClientDraws = clientDraws
        connection.setCursorRender(clientDraws: clientDraws)
    }
    /// Flip the mouse model with the atomic release/re-engage swap; `reappearAt` (host video
    /// px — the M3 hand-back position) warps the local pointer so leaving relative lands the
    /// cursor exactly where the host last had it.
    private func setDesktopMouse(_ on: Bool, reappearAt: (x: Int32, y: Int32)?) {
        guard desktopMouse != on else { return }
        let wasCaptured = captured
        if wasCaptured { releaseCapture() }
        desktopMouse = on
        if wasCaptured { engageCapture(fromClick: false) }
        window?.invalidateCursorRects(for: self)
        if on, let p = reappearAt, let sp = cgScreenPoint(forHostX: p.x, p.y) {
            CGWarpMouseCursorPosition(sp)
        }
        reconcileCursorRender()
    }
    /// The single cursor pull thread (both planes share the connection's cursor lock):
    /// latest-wins state at a short timeout + a non-blocking shape poll per iteration.
    /// Exits when the connection closes; events hop to main where all cursor state lives.
    private func startCursorPump(_ connection: PunktfunkConnection) {
        let thread = Thread { [weak self] in
            while true {
                do {
                    var newest: PunktfunkConnection.CursorStateEvent?
                    if let st = try connection.nextCursorState(timeoutMs: 100) {
                        newest = st
                        while let more = try connection.nextCursorState(timeoutMs: 0) {
                            newest = more // drain — latest wins
                        }
                    }
                    while let shape = try connection.nextCursorShape(timeoutMs: 0) {
                        DispatchQueue.main.async { self?.applyCursorShape(shape) }
                    }
                    if let st = newest {
                        DispatchQueue.main.async { self?.applyCursorState(st) }
                    }
                } catch {
                    return // connection closed — the session is over
                }
                if self == nil { return }
            }
        }
        thread.name = "pf-cursor-pump"
        thread.start()
    }
    private func applyCursorShape(_ ev: PunktfunkConnection.CursorShapeEvent) {
        guard let shape = Self.makeShape(ev) else {
            streamInputLog.warning("cursor shape rejected (\(ev.width)x\(ev.height)) — keeping the previous cursor")
            return
        }
        if hostCursors.count >= 64 { hostCursors.removeAll() } // degenerate host: reset
        hostCursors[ev.serial] = shape
        if cursorState?.serial == ev.serial {
            window?.invalidateCursorRects(for: self)
        }
    }
    private func applyCursorState(_ ev: PunktfunkConnection.CursorStateEvent) {
        let prev = cursorState
        cursorState = ev
        if prev?.visible != ev.visible || prev?.serial != ev.serial {
            window?.invalidateCursorRects(for: self)
        }
        // M3 host-driven auto-flip is DISABLED: `relative_hint` is derived from host cursor
        // VISIBILITY, and Windows hides the pointer for ordinary desktop activity (clicking,
        // typing) — not just when a game grabs it. Acting on those transients flipped
        // desktop→capture→desktop, which warped the cursor to view-centre and flushed held
        // buttons (a spurious button-up ~200 ms into every press → broke window drags). Until
        // the host exposes a real pointer-LOCK signal (ClipCursor/raw-input, not visibility),
        // the mouse model is user-driven only (⌃⌥⇧M). The hint still rides the wire, unused.
        _ = (lastHint, hintOverride)
    }
    /// Decode a forwarded straight-alpha RGBA shape into a CGImage + hotspot. The on-screen SIZE is
    /// NOT baked in here — it is applied per-use in `scaledCursor` from the live video-fit scale, so
    /// the same shape re-fits across window resizes / retina moves without a re-forward.
    private static func makeShape(_ ev: PunktfunkConnection.CursorShapeEvent) -> HostCursorShape? {
        let (w, h) = (ev.width, ev.height)
        guard w > 0, h > 0, ev.rgba.count >= w * h * 4,
              let provider = CGDataProvider(data: ev.rgba as CFData),
              let cg = CGImage(
                  width: w, height: h, bitsPerComponent: 8, bitsPerPixel: 32,
                  bytesPerRow: w * 4, space: CGColorSpaceCreateDeviceRGB(),
                  bitmapInfo: CGBitmapInfo(rawValue: CGImageAlphaInfo.last.rawValue),
                  provider: provider, decode: nil, shouldInterpolate: false,
                  intent: .defaultIntent)
        else { return nil }
        return HostCursorShape(
            cg: cg, width: w, height: h,
            hotX: min(ev.hotX, w - 1), hotY: min(ev.hotY, h - 1))
    }
    /// Points-per-host-pixel: the exact factor the video frame is aspect-fit into the view (the same
    /// `AVMakeRect` fit `hostPoint`/`cgScreenPoint` use). The host forwards the pointer bitmap in host
    /// framebuffer pixels — the mode we drive is in the client's BACKING pixels, so on retina this is
    /// ~1/backingScale and the pointer lands at its TRUE size relative to the streamed desktop
    /// (crisp, 1:1 with the video) rather than the 2×-inflated pixel-as-points it used to be. Because
    /// the bitmap grows with the host's display scaling (96 px at 300% DPI), scaling by this is what
    /// keeps a high-DPI host from forwarding a giant pointer. Falls back to 1 before the first
    /// mode/layout.
    private func cursorFitScale() -> CGFloat {
        guard let connection else { return 1 }
        let mode = connection.currentMode()
        guard mode.width > 0, mode.height > 0, bounds.width > 0, bounds.height > 0 else { return 1 }
        let fit = AVMakeRect(
            aspectRatio: CGSize(width: Int(mode.width), height: Int(mode.height)), insideRect: bounds)
        guard fit.width > 0 else { return 1 }
        return fit.width / CGFloat(mode.width)
    }
    /// Build the `NSCursor` for a cached shape at the CURRENT video-fit scale (see `cursorFitScale`).
    /// Both the image size and the hotspot scale together so the click point stays true.
    private func scaledCursor(_ shape: HostCursorShape) -> NSCursor {
        let scale = cursorFitScale()
        let sw = max(1, (CGFloat(shape.width) * scale).rounded())
        let sh = max(1, (CGFloat(shape.height) * scale).rounded())
        let image = NSImage(cgImage: shape.cg, size: NSSize(width: sw, height: sh))
        let hot = NSPoint(
            x: min(CGFloat(shape.hotX) * scale, sw - 1),
            y: min(CGFloat(shape.hotY) * scale, sh - 1))
        return NSCursor(image: image, hotSpot: hot)
    }
    /// Host video px → CG GLOBAL screen coordinates (top-left origin, the
    /// `CGWarpMouseCursorPosition` convention `CursorCapture` established) through the
    /// aspect-fit letterbox — the inverse direction of `hostPoint(from:)`.
    private func cgScreenPoint(forHostX hx: Int32, _ hy: Int32) -> CGPoint? {
        guard let connection, let window else { return nil }
        let mode = connection.currentMode()
        guard mode.width > 0, mode.height > 0 else { return nil }
        let fit = AVMakeRect(
            aspectRatio: CGSize(width: Int(mode.width), height: Int(mode.height)),
            insideRect: bounds)
        guard fit.width > 0, fit.height > 0 else { return nil }
        let u = (CGFloat(hx) / CGFloat(mode.width)).clamped(to: 0...1)
        let v = (CGFloat(hy) / CGFloat(mode.height)).clamped(to: 0...1)
        let videoMinYTop = bounds.height - fit.maxY
        let pTop = CGPoint(x: fit.minX + u * fit.width, y: videoMinYTop + v * fit.height)
        let inView = CGPoint(x: pTop.x, y: bounds.height - pTop.y)
        let inWindow = convert(inView, to: nil)
        let onScreen = window.convertPoint(toScreen: inWindow)
        let primaryHeight = NSScreen.screens.first?.frame.height ?? 0
        return CGPoint(x: onScreen.x, y: primaryHeight - onScreen.y)
    }
    /// A single local monitor for motion + buttons, installed only while captured. A local
    /// monitor is more robust than view overrides for relative motion: it sidesteps the
    /// `window.acceptsMouseMovedEvents`/tracking-area/responder-chain requirements, and
@@ -647,12 +842,10 @@ public final class StreamLayerView: NSView {
                streamInputLog.info("mouse-mode chord ignored: gamescope host is relative-only")
                return
            }
-            let wasCaptured = self.captured
+            // A manual flip outranks the standing host hint until the hint next CHANGES.
-            if wasCaptured { self.releaseCapture() }
+            self.hintOverride = true
-            self.desktopMouse.toggle()
+            self.setDesktopMouse(!self.desktopMouse, reappearAt: nil)
-            if wasCaptured { self.engageCapture(fromClick: false) }
+            streamInputLog.info("chord: mouse mode \(self.desktopMouse ? "desktop" : "capture", privacy: .public)")
            self.window?.invalidateCursorRects(for: self)
            streamInputLog.info("mouse mode: \(self.desktopMouse ? "desktop" : "capture", privacy: .public)")
        }
        // The cross-client combos (⌃⌥⇧Q/D/S — Ctrl+Alt+Shift on the other clients), delivered by
        // the monitor only while captured; the same key-window ownership rule as ⌘⎋ throughout.
@@ -692,6 +885,14 @@ public final class StreamLayerView: NSView {
        if mode == .desktop && !absOK {
            streamInputLog.info("desktop mouse mode unavailable on a gamescope host (relative-only) — using capture")
        }
        // Cursor channel (M2): the host stopped compositing the pointer — drain its shape/
        // state planes and draw the pointer as the real NSCursor (plus the M3 auto-flip).
        if connection.hostSupportsCursor {
            cursorChannelActive = true
            streamInputLog.info("cursor channel negotiated — host cursor renders locally")
            startCursorPump(connection)
            reconcileCursorRender() // initial render mode (a capture-model start composites)
        }
        // Presenter choice + lifecycle live in SessionPresenter (shared with iOS/tvOS): stage-2
        // (explicit VTDecompressionSession decode + a CAMetalLayer/display-link present) by
@@ -751,6 +952,11 @@ public final class StreamLayerView: NSView {
            matchFollower?.noteSize(
                widthPx: Int(px.width.rounded()), heightPx: Int(px.height.rounded()))
        }
        // The video-fit scale just changed (resize / retina move); rebuild the worn host pointer at
        // the new scale so it tracks the video instead of freezing at its build-time size.
        if captured, desktopMouse, cursorChannelActive {
            window?.invalidateCursorRects(for: self)
        }
    }
    public override func viewDidChangeBackingProperties() {
@@ -781,6 +987,14 @@ public final class StreamLayerView: NSView {
        matchFollower = nil
        lastDecodedContentSize = nil // the next session re-derives it from its first frame
        connection = nil
        // Cursor-channel state is per-session: without this reset a next session against a
        // host WITHOUT the cap would wear this session's stale shapes (`cursorChannelActive`
        // stayed latched true across sessions).
        cursorChannelActive = false
        cursorState = nil
        hostCursors.removeAll()
        sentClientDraws = nil
        window?.invalidateCursorRects(for: self)
    }
    deinit {
@@ -333,6 +333,13 @@ public final class StreamViewController: StreamViewControllerBase {
            guard self?.captureEnabled == true else { return }
            connection?.send(event)
        }
        // Apple Pencil → the stylus plane, only against a pen-capable host (elsewhere the
        // Pencil stays a finger, exactly as before). Same trust gate as touch.
        streamView.penEnabled = connection.hostSupportsPen
        streamView.onPenBatch = { [weak self, weak connection] batch in
            guard self?.captureEnabled == true else { return }
            connection?.sendPen(batch)
        }
        // Indirect pointer (mouse/trackpad) WITHOUT a lock → absolute cursor + buttons + scroll.
        // While the scene is pointer-LOCKED the GCMouse path owns motion AND buttons AND scroll, so
        // the whole UIKit indirect path is gated off here (`gcMouseForwarding`). The trackpad and a
@@ -499,6 +506,8 @@ public final class StreamViewController: StreamViewControllerBase {
        // onTouchEvent can still deliver the button-up.
        streamView.resetTouchInput()
        streamView.onTouchEvent = nil
        streamView.onPenBatch = nil // after reset — the pen's leave-range sample rides it
        streamView.penEnabled = false
        streamView.onPointerMoveAbs = nil
        streamView.onPointerButton = nil
        streamView.onScroll = nil
@@ -692,6 +701,12 @@ final class StreamLayerUIView: UIView {
    /// Direct fingers / Pencil → wire events: real touches in passthrough mode, or the
    /// touch-driven mouse events (`TouchMouse`) in the trackpad/pointer modes.
    var onTouchEvent: ((PunktfunkInputEvent) -> Void)?
    /// Apple Pencil → state-full pen sample batches (the stylus plane). Active only while
    /// `penEnabled`; without it the Pencil stays on the finger path exactly as before.
    var onPenBatch: (([PunktfunkPenSample]) -> Void)?
    /// The host advertised `HOST_CAP_PEN`, so Pencil input splits out of the finger path onto
    /// the pen plane — independent of the touch-input mode (drawing must not depend on it).
    var penEnabled = false
    /// Indirect pointer (mouse/trackpad with no lock) → absolute cursor moves.
    var onPointerMoveAbs: ((HostPoint) -> Void)?
    /// Indirect-pointer buttons (GameStream ids: 1=left 3=right); `down` = press.
@@ -715,10 +730,21 @@ final class StreamLayerUIView: UIView {
    /// The finger route latched at gesture start — a Settings change mid-gesture applies to
    /// the NEXT touch, so one gesture never splits across input models.
    private var fingerRoute: TouchInputMode?
    /// The Apple Pencil pipeline (contacts + hover + squeeze/tap → pen samples).
    private lazy var pencil: PencilStream = {
        let stream = PencilStream()
        stream.send = { [weak self] batch in self?.onPenBatch?(batch) }
        stream.videoNorm = { [weak self] point in
            guard let h = self?.hostPoint(from: point) else { return nil }
            return (Float(h.x) / Float(max(h.w - 1, 1)), Float(h.y) / Float(max(h.h - 1, 1)))
        }
        return stream
    }()
    /// Release anything the touch-driven mouse holds and forget gesture state — session stop.
    func resetTouchInput() {
        touchMouse.reset()
        pencil.reset() // leaves range → the host lifts anything still inked
        fingerRoute = nil
        setSoftKeyboardVisible(false) // a stream that's gone takes its keyboard with it
    }
@@ -755,6 +781,11 @@ final class StreamLayerUIView: UIView {
        scrollPan.allowedScrollTypesMask = .all
        scrollPan.allowedTouchTypes = []
        addGestureRecognizer(scrollPan)
        // Pencil squeeze / double-tap → the pen plane's barrel buttons (no-op while
        // `penEnabled` is false — PencilStream ignores interactions out of range).
        let pencilInteraction = UIPencilInteraction()
        pencilInteraction.delegate = pencil
        addInteraction(pencilInteraction)
        #endif
        backgroundColor = .black
    }
@@ -779,17 +810,32 @@ final class StreamLayerUIView: UIView {
    private enum TouchKind { case down, move, up, cancel }
    /// Split a touch batch by kind: an INDIRECT POINTER (mouse/trackpad with no lock) drives
-    /// the host cursor as an absolute mouse; everything else (direct finger, Pencil) is a host
+    /// the host cursor as an absolute mouse; a Pencil goes to the pen plane when the host
-    /// touch. Mixed batches are possible, so partition rather than branch on the first touch.
+    /// supports it; everything else (direct finger — and the Pencil toward a pen-less host)
    /// is a host touch. Mixed batches are possible, so partition rather than branch on the
    /// first touch.
    private func route(_ touches: Set<UITouch>, event: UIEvent?, kind: TouchKind) {
        var fingers: Set<UITouch> = []
        var pencilTouches: Set<UITouch> = []
        for touch in touches {
            if touch.type == .indirectPointer {
                handleIndirectPointer(touch, event: event, kind: kind)
            } else if penEnabled, touch.type == .pencil {
                pencilTouches.insert(touch)
            } else {
                fingers.insert(touch)
            }
        }
        if !pencilTouches.isEmpty {
            let phase: PencilStream.Phase =
                switch kind {
                case .down: .down
                case .move: .move
                case .up: .up
                case .cancel: .cancel
                }
            pencil.touches(pencilTouches, event: event, phase: phase, in: self)
        }
        if !fingers.isEmpty { forwardFingers(fingers, kind: kind) }
    }
@@ -861,8 +907,11 @@ final class StreamLayerUIView: UIView {
        }
    }
-    /// Button-less mouse/trackpad movement (no lock) → absolute cursor move.
+    /// Button-less mouse/trackpad movement (no lock) → absolute cursor move — unless it is a
    /// hovering PENCIL (`zOffset > 0`) on a pen-capable host, which becomes in-range pen
    /// samples (hover preview with distance/tilt/azimuth) instead of a cursor move.
    @objc private func handleHover(_ recognizer: UIHoverGestureRecognizer) {
        if penEnabled, pencil.maybeHover(recognizer, in: self) { return }
        switch recognizer.state {
        case .began, .changed:
            if let h = hostPoint(from: recognizer.location(in: self)) { onPointerMoveAbs?(h) }
@@ -10,9 +10,17 @@ import {
  showModal,
  staticClasses,
 } from "@decky/ui";
-import { definePlugin, routerHook } from "@decky/api";
+import { definePlugin, routerHook, toaster } from "@decky/api";
 import { FC } from "react";
-import { FaDownload, FaLock, FaLockOpen, FaPlay, FaSyncAlt, FaTv } from "react-icons/fa";
+import {
  FaDownload,
  FaLock,
  FaLockOpen,
  FaPlay,
  FaPlus,
  FaSyncAlt,
  FaTv,
 } from "react-icons/fa";
 import { PluginErrorBoundary } from "./boundary";
 import {
  applyUpdate,
@@ -31,7 +39,19 @@ import {
 import { streamPin } from "./library";
 import { PunktfunkRoute, ROUTE } from "./page";
 import { PairModal } from "./pair";
-import { ensureGamepadUiShortcut } from "./steam";
+import { ensureGamepadUiShortcut, recreateShortcuts } from "./steam";
 // Recovery action for "the Punktfunk library entry vanished" — recreates the visible shortcut.
 // Deleting the shortcut (optionally + reinstalling the plugin) leaves a stale appId in Steam's
 // CEF localStorage that self-heal fixes on the next mount, but this gives an in-session button
 // that works even without a reload. Always ends in a toast so the tap has feedback.
 async function recreatePunktfunkShortcut(): Promise<void> {
  const appId = await recreateShortcuts();
  toaster.toast({
    title: "Punktfunk",
    body: appId != null ? "Shortcut restored to your library" : "Couldn't create the shortcut",
  });
 }
 // ----------------------------------------------------------------------------------------
 // QAM panel — quick status + entry into the full page + one-tap stream for known hosts
@@ -190,6 +210,16 @@ const QamPanel: FC = () => {
            {checking ? "Checking…" : "Check for updates"}
          </ButtonItem>
        </PanelSectionRow>
        <PanelSectionRow>
          <ButtonItem
            layout="below"
            description="Missing the Punktfunk entry in your library? This puts it back."
            onClick={() => void recreatePunktfunkShortcut()}
          >
            <FaPlus style={{ marginRight: "0.5em" }} />
            Recreate library shortcut
          </ButtonItem>
        </PanelSectionRow>
      </PanelSection>
    </>
  );
@@ -55,6 +55,29 @@ declare const collectionStore:
  | { SetAppsAsHidden?: (appIds: number[], hidden: boolean) => void }
  | undefined;
 // SteamUI's appStore indexes every registered app/shortcut by appId; a remembered appId whose
 // overview is gone was deleted out from under us (the user removed the library entry). We must
 // verify this because the remembered appId lives in Steam's CEF localStorage — which survives a
 // plugin UNINSTALL/REINSTALL — so a manually-deleted shortcut otherwise leaves a dangling appId
 // that the reuse path below silently repoints (SetShortcut* on a dead id is a no-op), and the
 // entry never comes back.
 declare const appStore:
  | { GetAppOverviewByAppID?: (appId: number) => unknown | null }
  | undefined;
 /** True if a remembered appId still maps to a live Steam shortcut. When appStore is unavailable
 *  we can't tell, so assume it exists — better to keep reusing than risk a duplicate library
 *  entry from a false "missing". A confident null means the shortcut was deleted → recreate. */
 function shortcutStillExists(appId: number): boolean {
  try {
    const get = appStore?.GetAppOverviewByAppID;
    if (!get) return true; // no way to verify — preserve the reuse path
    return get(appId) != null;
  } catch {
    return true;
  }
 }
 /** Set a shortcut's library visibility (best-effort, deferred — the overview registers a moment
 *  after AddShortcut). Hides the stateful stream shortcut; keeps the gamepad-UI one visible. */
 function setShortcutHidden(appId: number, hidden: boolean): void {
@@ -199,8 +222,10 @@ async function ensureStreamShortcut(): Promise<{ appId: number; runner: string }
  const startDir = info.runner.replace(/\/[^/]*$/, ""); // the plugin's bin/ dir
  void ensureControllerConfig(); // fire-and-forget — never blocks the launch
  // Reuse the remembered shortcut only if it still exists — a stale appId (shortcut deleted, key
  // outlived it across a reinstall) must fall through to AddShortcut, not be silently repointed.
  const remembered = recall(STORAGE_KEY_STREAM);
-  if (remembered != null) {
+  if (remembered != null && shortcutStillExists(remembered)) {
    SteamClient.Apps.SetShortcutExe(remembered, SHELL);
    SteamClient.Apps.SetShortcutStartDir(remembered, startDir);
    SteamClient.Apps.SetShortcutName(remembered, SHORTCUT_NAME);
@@ -236,8 +261,11 @@ export async function ensureGamepadUiShortcut(): Promise<number | null> {
    // home). %command% expands to the shortcut exe (/bin/sh); the wrapper rides behind as an arg.
    const launchOpts = `PF_BROWSE=1 %command% "${info.runner}"`;
    // Reuse the remembered entry only if it still exists; a stale appId (deleted shortcut whose
    // localStorage key survived a plugin reinstall) falls through to AddShortcut so the visible
    // library entry actually comes back instead of repointing a dead id.
    let appId = recall(STORAGE_KEY_UI);
-    if (appId != null) {
+    if (appId != null && shortcutStillExists(appId)) {
      SteamClient.Apps.SetShortcutExe(appId, SHELL);
      SteamClient.Apps.SetShortcutStartDir(appId, startDir);
      SteamClient.Apps.SetShortcutName(appId, SHORTCUT_NAME);
@@ -256,6 +284,30 @@ export async function ensureGamepadUiShortcut(): Promise<number | null> {
  }
 }
 /**
 * Force the visible "Punktfunk" library entry back into existence — the recovery button for
 * "my shortcut disappeared". Drops any remembered appId that no longer maps to a live shortcut
 * (so it can't shadow a fresh AddShortcut), then re-ensures. Safe to press anytime: a shortcut
 * that still exists is left in place (no duplicate); a missing one is recreated. Covers the case
 * self-heal-on-mount can't — deleting the shortcut WITHOUT reinstalling (no mount → no ensure).
 * Returns the (new or existing) visible appId, or null on failure.
 */
 export async function recreateShortcuts(): Promise<number | null> {
  for (const key of [STORAGE_KEY_STREAM, STORAGE_KEY_UI]) {
    const id = recall(key);
    if (id != null && !shortcutStillExists(id)) {
      try {
        localStorage.removeItem(artKey(id)); // stale art marker for the dead appId
        localStorage.removeItem(key);
      } catch {
        /* ignore */
      }
    }
  }
  // Recreate the visible entry now; the hidden stream shortcut re-registers lazily on next launch.
  return ensureGamepadUiShortcut();
 }
 /** Launch the stateless gamepad-UI shortcut (console home) from the plugin, e.g. a QAM button. */
 export async function launchGamepadUi(): Promise<void> {
  const appId = await ensureGamepadUiShortcut();
@@ -547,6 +547,7 @@ impl AppModel {
                    crate::video::decodable_codecs(), // codecs (unused by the probe, but honest)
                    0,                                // preferred_codec: no preference
                    None, // display_hdr: probe connect, nothing presents
                    0,    // client_caps: probe connect, nothing renders a cursor
                    None, // launch: probe connect, no game
                    pin,
                    Some(identity),
@@ -523,6 +523,9 @@ async fn session(args: Args) -> Result<()> {
            // writes it into the virtual display's EDID (CTA HDR block), so the EDID-forwarding
            // path can be validated headlessly (check the host's monitor caps / ADD log line).
            display_hdr: punktfunk_core::client::display_hdr_env_override(),
            // No CLIENT_CAP_CURSOR: this headless tool renders nothing — advertising it would
            // just strip the pointer from the dumped bitstream.
            client_caps: 0,
        }
        .encode(),
    )
@@ -172,6 +172,11 @@ mod session_main {
            // defaults for Linux clients; `PUNKTFUNK_CLIENT_PEAK_NITS` (read in the session
            // pump) pins one manually.
            display_hdr: None,
            // The presenter renders the host cursor locally in desktop mouse mode (M2 cursor
            // channel); capture-mode sessions keep the composited cursor, so only advertise
            // when the session STARTS in desktop mode. The host gates further (Linux portal
            // compositors only).
            cursor_forward: settings.mouse_mode() == trust::MouseMode::Desktop,
            mic_enabled: settings.mic_enabled,
            clipboard,
            // The Settings preference (auto → VAAPI where it exists; the presenter
@@ -56,6 +56,7 @@ pub fn run_speed_probe(
        decodable_codecs(),
        0,    // preferred_codec: no preference
        None, // display_hdr: probe connect, nothing presents
        0,    // client_caps: probe connect, nothing renders a cursor
        None, // launch: no game
        pin,
        Some(identity),
@@ -30,6 +30,12 @@ pipewire = "0.9"
 libc = "0.2"
 # ashpd 0.13 uses the tokio runtime for the one-time portal handshake (control plane).
 tokio = { version = "1", features = ["rt", "rt-multi-thread", "net", "time"] }
 # XFixes cursor source for gamescope (remote-desktop-sweep Phase C): gamescope paints no
 # `SPA_META_Cursor`, so the pointer never reaches the PipeWire node. We read the shape/hotspot/
 # visibility from gamescope's nested Xwayland via XFixes instead and feed the existing cursor slot.
 # `RustConnection` is the pure-Rust default (no libxcb link → no new C dependency on the host); the
 # `xfixes` feature (auto-pulls `render` + `shape`) is what exposes GetCursorImage/SelectCursorInput.
 x11rb = { version = "0.13", default-features = false, features = ["xfixes"] }
 [target.'cfg(target_os = "windows")'.dependencies]
 # The host<->driver wire contract for the sealed frame channel (control IOCTL structs + frame header).
@@ -43,7 +49,9 @@ windows = { version = "0.62", features = [
    "Win32_Graphics_Direct3D_Fxc",
    "Win32_Graphics_Dxgi",
    "Win32_Graphics_Dxgi_Common",
    "Win32_Graphics_Gdi",
    "Win32_System_LibraryLoader",
    "Win32_System_StationsAndDesktops",
    "Win32_System_Memory",
    "Win32_System_Threading",
    "Win32_UI_HiDpi",
@@ -69,6 +69,34 @@ pub trait Capturer: Send {
    /// SDR / a backend that doesn't expose it (the default — Linux capture has no HDR path yet).
    /// The stream loop forwards this to the encoder (in-band SEI) and the client (`0xCE` datagram),
    /// so the two stay a single source of truth. May change mid-session if the source is regraded.
    /// The capture source's LIVE cursor state, when it arrives out-of-band from the frames
    /// (the Windows IddCx hardware-cursor channel). Polled by the encode loop every tick and
    /// preferred over `CapturedFrame::cursor` — with a hardware cursor, pointer-only moves
    /// produce NO new frame, so the frame-attached overlay would go stale on a static desktop.
    /// Default `None`: the Linux portal path attaches its cursor to frames instead.
    fn cursor(&mut self) -> Option<pf_frame::CursorOverlay> {
        None
    }
    /// LIVE cursor-render flip for a cursor-forward session (design/remote-desktop-sweep.md §8):
    /// `on = true` — the client draws the pointer, keep it OUT of the video; `on = false` —
    /// the capture mouse model, the pointer must be IN the video again. The Windows IDD
    /// capturer implements the composite side ITSELF (slot-copy + alpha-blended quad from the
    /// GDI poller) — a declared IddCx hardware cursor is irrevocable, so DWM can never be
    /// handed the job back. Called every encode tick (implementations cache; steady state is
    /// one compare). Default no-op: the Linux portal never bakes the pointer into frames —
    /// the encode loop blends its overlay instead.
    fn set_cursor_forward(&mut self, _on: bool) {}
    /// Attach a gamescope cursor source (remote-desktop-sweep Phase C). gamescope paints no
    /// `SPA_META_Cursor`, so [`cursor`](Self::cursor)'s slot stays empty — this hands the Linux
    /// portal capturer gamescope's nested Xwayland `(DISPLAY, XAUTHORITY)` targets (it may run
    /// several — one per `--xwayland-count`) so it reads the pointer shape/position over X11
    /// (XFixes + QueryPointer), following whichever display is focused, and publishes it into that
    /// same slot. Called once, after the capturer is built, only for gamescope sessions. Default
    /// no-op: every non-gamescope capturer already has a cursor source.
    fn attach_gamescope_cursor(&mut self, _targets: Vec<(String, Option<String>)>) {}
    fn hdr_meta(&self) -> Option<punktfunk_core::quic::HdrMeta> {
        None
    }
@@ -367,6 +395,26 @@ pub type FrameChannelSender = std::sync::Arc<
    dyn Fn(&pf_driver_proto::control::SetFrameChannelRequest) -> Result<()> + Send + Sync,
 >;
 /// Delivery closure for the v5 hardware-cursor channel (`IOCTL_SET_CURSOR_CHANNEL`) — same
 /// facade contract as [`FrameChannelSender`]. `Some` also OPTS THE SESSION IN: the capturer
 /// creates + delivers the cursor section only when the host hands it a sender (the negotiated
 /// cursor-forward sessions), and the driver only declares the hardware cursor once that
 /// delivery lands — so a plain session keeps DWM's composited pointer untouched.
 #[cfg(target_os = "windows")]
 pub type CursorChannelSender = std::sync::Arc<
    dyn Fn(&pf_driver_proto::control::SetCursorChannelRequest) -> Result<()> + Send + Sync,
 >;
 /// The mid-stream cursor-render flip (`IOCTL_SET_CURSOR_FORWARD`, proto v6) as a host-facade
 /// closure — same contract as [`CursorChannelSender`]. `bool` = declare the IddCx hardware
 /// cursor (`true`) or stand it down (`false`; the host facade additionally forces the same-mode
 /// re-commit that actualises the OS's software-cursor default). The capturer drives this from
 /// its secure-desktop watch: UAC/Winlogon render only through the software-cursor path, so a
 /// path pinned to the hardware cursor never presents them (the 0.18.0 secure-desktop
 /// regression).
 #[cfg(target_os = "windows")]
 pub type CursorForwardSender = std::sync::Arc<dyn Fn(bool) -> Result<()> + Send + Sync>;
 // One-time PipeWire library init, shared by the video (portal) and audio capture threads.
 #[cfg(target_os = "linux")]
 pub mod pwinit;
@@ -424,6 +472,7 @@ pub fn open_virtual_output(
    allow_zerocopy: bool,
    want_444: bool,
    policy: ZeroCopyPolicy,
    expect_exact_dims: bool,
 ) -> Result<Box<dyn Capturer>> {
    linux::PortalCapturer::from_virtual_output(
        remote_fd,
@@ -433,6 +482,7 @@ pub fn open_virtual_output(
        allow_zerocopy,
        want_444,
        policy,
        expect_exact_dims,
    )
    .map(|c| Box::new(c) as Box<dyn Capturer>)
 }
@@ -450,6 +500,8 @@ pub fn open_idd_push(
    pyrowave: bool,
    keepalive: Box<dyn Send>,
    sender: FrameChannelSender,
    cursor_sender: Option<CursorChannelSender>,
    cursor_forward: Option<CursorForwardSender>,
 ) -> std::result::Result<Box<dyn Capturer>, (anyhow::Error, Box<dyn Send>)> {
    idd_push::IddPushCapturer::open(
        target,
@@ -459,6 +511,8 @@ pub fn open_idd_push(
        pyrowave,
        keepalive,
        sender,
        cursor_sender,
        cursor_forward,
    )
    .map(|c| Box::new(c) as Box<dyn Capturer>)
 }
@@ -22,6 +22,10 @@
 use super::{CapturedFrame, Capturer, DmabufFrame, FramePayload, PixelFormat, ZeroCopyPolicy};
 use anyhow::{anyhow, Context, Result};
 // gamescope cursor source (remote-desktop-sweep Phase C) — feeds `cursor_live` from XFixes when
 // the PipeWire node carries no `SPA_META_Cursor` (gamescope's does not).
 mod xfixes_cursor;
 use std::os::fd::OwnedFd;
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::mpsc::{sync_channel, Receiver, RecvTimeoutError, TryRecvError};
@@ -57,6 +61,12 @@ pub struct PortalCapturer {
    /// renegotiation before declaring the source lost. Cleared whenever a frame arrives or the stream
    /// is `Streaming`.
    stall_since: Option<std::time::Instant>,
    /// The LIVE cursor overlay, published by the PipeWire thread from every buffer's
    /// `SPA_META_Cursor` — including the cursor-only "corrupted" buffers that never become
    /// frames. [`Capturer::cursor`] serves it so the encode loop's forwarder tracks pointer-only
    /// motion on a static desktop; the frame-attached overlay alone goes stale between damage
    /// frames (the same gap the Windows IddCx channel fills, and why the tick prefers LIVE).
    cursor_live: Arc<std::sync::Mutex<Option<pf_frame::CursorOverlay>>>,
    /// True when this capture runs the VAAPI dmabuf passthrough (a LINEAR-dmabuf-only offer). If
    /// that offer never negotiates, [`next_frame`](Capturer::next_frame)'s timeout branch latches
    /// the process-wide downgrade ([`pf_zerocopy::note_vaapi_dmabuf_failed`]) so the pipeline
@@ -82,6 +92,12 @@ pub struct PortalCapturer {
    /// is, releasing the compositor-side output via the keepalive's own `Drop`. `None` for the
    /// portal source (its session ends with the portal thread's zbus connection).
    _keepalive: Option<Box<dyn Send>>,
    /// The gamescope XFixes cursor reader (remote-desktop-sweep Phase C), when this capturer
    /// serves a gamescope node. `Some` after
    /// [`attach_gamescope_cursor`](Capturer::attach_gamescope_cursor); its `Drop` stops the reader
    /// thread, so it lives exactly as long as the capturer. `None` on the portal path (its cursor
    /// comes from `SPA_META_Cursor`).
    _gs_cursor: Option<xfixes_cursor::XFixesCursorSource>,
 }
 impl PortalCapturer {
@@ -114,10 +130,17 @@ impl PortalCapturer {
            "ScreenCast portal session started; connecting PipeWire"
        );
        // This portal path (GameStream / monitor capture) is always 4:2:0, so allow zero-copy as before.
-        Ok(
+        Ok(spawn_pipewire(
-            spawn_pipewire(Some(fd), node_id, None, true, false, want_hdr, policy)?
+            Some(fd),
-                .into_capturer(node_id, None),
+            node_id,
-        )
+            None,
            true,
            false,
            want_hdr,
            policy,
            false,
        )?
        .into_capturer(node_id, None))
    }
    /// Build a capturer from an already-created virtual output's PipeWire node. The host facade
@@ -137,11 +160,13 @@ impl PortalCapturer {
        allow_zerocopy: bool,
        want_444: bool,
        policy: ZeroCopyPolicy,
        expect_exact_dims: bool,
    ) -> Result<PortalCapturer> {
        tracing::info!(
            node_id,
            allow_zerocopy,
            want_444,
            expect_exact_dims,
            "connecting PipeWire to virtual output"
        );
        // Virtual outputs are SDR-only upstream (Mutter's RecordVirtual streams advertise 8-bit
@@ -154,6 +179,7 @@ impl PortalCapturer {
            want_444,
            false,
            policy,
            expect_exact_dims,
        )?
        .into_capturer(node_id, Some(keepalive)))
    }
@@ -176,6 +202,8 @@ struct PwHandles {
    hdr_offer: bool,
    /// See [`PortalCapturer::hdr_negotiated`].
    hdr_negotiated: Arc<AtomicBool>,
    /// See [`PortalCapturer::cursor_live`].
    cursor_live: Arc<std::sync::Mutex<Option<pf_frame::CursorOverlay>>>,
    quit: ::pipewire::channel::Sender<()>,
    join: thread::JoinHandle<()>,
 }
@@ -195,10 +223,12 @@ impl PwHandles {
            vaapi_dmabuf: self.vaapi_dmabuf,
            hdr_offer: self.hdr_offer,
            hdr_negotiated: self.hdr_negotiated,
            cursor_live: self.cursor_live,
            node_id,
            quit: Some(self.quit),
            join: Some(self.join),
            _keepalive: keepalive,
            _gs_cursor: None,
        }
    }
 }
@@ -206,6 +236,7 @@ impl PwHandles {
 /// Spawn the PipeWire consumer thread for `node_id` (fd `Some` = portal remote, `None` =
 /// default daemon) and return its [`PwHandles`]. `preferred` seeds the format negotiation's
 /// default size/framerate — for Mutter virtual monitors this is what actually sizes the monitor.
 #[allow(clippy::too_many_arguments)]
 fn spawn_pipewire(
    fd: Option<OwnedFd>,
    node_id: u32,
@@ -224,6 +255,12 @@ fn spawn_pipewire(
    // Encode-backend facts resolved by the facade (never re-derived here) — the one-way
    // capture→encode edge (plan §W6).
    policy: ZeroCopyPolicy,
    // The producer's FIRST negotiation is for a sacrificial mode and a renegotiation to
    // `preferred`'s dims is guaranteed to follow (KWin virtual outputs — see kwin.rs `create`):
    // skip whole buffers until the negotiated size matches, so the pipeline never builds against
    // the doomed birth mode. `false` everywhere else (Mutter SIZES the monitor from negotiation,
    // gamescope fixates its own — gating those would starve legitimate first frames).
    expect_exact_dims: bool,
 ) -> Result<PwHandles> {
    // Frames flow from the pipewire thread over a small bounded channel.
    let (frame_tx, frame_rx) = sync_channel::<CapturedFrame>(8);
@@ -237,6 +274,8 @@ fn spawn_pipewire(
    let broken_cb = broken.clone();
    let hdr_negotiated = Arc::new(AtomicBool::new(false));
    let hdr_negotiated_cb = hdr_negotiated.clone();
    let cursor_live = Arc::new(std::sync::Mutex::new(None::<pf_frame::CursorOverlay>));
    let cursor_live_cb = cursor_live.clone();
    // pipewire's own cross-thread channel: the receiver attaches to the loop and quits it; the
    // sender lives on the capturer and fires in its `Drop`. Absolute `::pipewire` path — the
    // inner `mod pipewire` shadows the crate name at this scope.
@@ -272,12 +311,14 @@ fn spawn_pipewire(
                streaming_cb,
                broken_cb,
                hdr_negotiated_cb,
                cursor_live_cb,
                zerocopy,
                want_444,
                want_hdr,
                preferred,
                quit_rx,
                policy,
                expect_exact_dims,
            ) {
                tracing::error!(error = %format!("{e:#}"), "pipewire capture thread failed");
            }
@@ -292,6 +333,7 @@ fn spawn_pipewire(
        vaapi_dmabuf,
        hdr_offer: want_hdr,
        hdr_negotiated,
        cursor_live,
        quit: quit_tx,
        join,
    })
@@ -302,6 +344,24 @@ impl Capturer for PortalCapturer {
        self.frame_within(Duration::from_secs(10))
    }
    fn cursor(&mut self) -> Option<pf_frame::CursorOverlay> {
        // The PipeWire thread's live cursor slot (fed by every buffer's meta, frames or not) —
        // lets the forwarder track pointer-only motion on a static desktop. See `cursor_live`.
        // On a gamescope node the meta never arrives; the XFixes source (attached below) fills
        // the same slot instead.
        self.cursor_live.lock().ok().and_then(|slot| slot.clone())
    }
    fn attach_gamescope_cursor(&mut self, targets: Vec<(String, Option<String>)>) {
        // gamescope paints no `SPA_META_Cursor`, so `cursor_live` would stay empty. Spawn the
        // XFixes reader to publish gamescope's pointer into that SAME slot — `cursor()` above then
        // serves it and the encode loop composites it, exactly like the portal path. It connects
        // to every nested Xwayland and follows the focused one's pointer. A failure (no Xwayland /
        // no XFixes) logs and leaves the slot empty = today's cursorless gamescope.
        self._gs_cursor =
            xfixes_cursor::XFixesCursorSource::spawn(targets, Arc::clone(&self.cursor_live));
    }
    fn next_frame_within(&mut self, budget: Duration) -> Result<CapturedFrame> {
        self.frame_within(budget)
    }
@@ -868,13 +928,21 @@ mod pipewire {
        /// Bumps whenever the bitmap (`rgba`/`bw`/`bh`) changes — stable across position-only moves,
        /// so the GPU encoder re-uploads its cursor texture only on change.
        serial: u64,
        /// The compositor-reported hotspot — carried on the overlay for the cursor-forward
        /// channel (the blend path uses the pre-adjusted `x`/`y` and never reads it).
        hot_x: i32,
        hot_y: i32,
    }
    impl CursorState {
-        /// A shareable overlay for the GPU encode paths (blended at encode time), or `None` when
+        /// A shareable overlay for the encode/forward paths, or `None` before the first bitmap
-        /// there is nothing to draw. Cheap: clones an `Arc` + a few scalars.
+        /// arrived. A HIDDEN pointer still yields `Some` (with `visible: false`): the
        /// cursor-forward channel needs "known but hidden" — an app grabbed the pointer, the
        /// client's relative-mode hint (M3) — which is a different fact from "no cursor yet".
        /// The encode loop strips invisible overlays before any blend path sees the frame.
        /// Cheap: clones an `Arc` + a few scalars.
        fn overlay(&self) -> Option<pf_frame::CursorOverlay> {
-            if !self.visible || self.rgba.is_empty() {
+            if self.rgba.is_empty() {
                return None;
            }
            Some(pf_frame::CursorOverlay {
@@ -884,6 +952,9 @@ mod pipewire {
                h: self.bh,
                rgba: self.rgba.clone(),
                serial: self.serial,
                hot_x: self.hot_x.max(0) as u32,
                hot_y: self.hot_y.max(0) as u32,
                visible: self.visible,
            })
        }
    }
@@ -934,6 +1005,21 @@ mod pipewire {
        dbg_log_n: u64,
        /// Cursor-as-metadata state, composited into the CPU de-pad path (see `consume_frame`).
        cursor: CursorState,
        /// LIVE overlay slot shared with [`super::PortalCapturer::cursor_live`] — refreshed after
        /// every `update_cursor_meta`, including from cursor-only buffers that never become frames.
        cursor_live: Arc<std::sync::Mutex<Option<pf_frame::CursorOverlay>>>,
        /// `Some((w, h))` while the producer's negotiated size is a sacrificial birth mode and a
        /// renegotiation to these dims is guaranteed (KWin virtual outputs — kwin.rs `create`):
        /// `.process` skips whole buffers until the negotiated size matches, then clears this
        /// (self-disarming — later legitimate resizes are unaffected). `None` = no gating.
        expect_dims: Option<(u32, u32)>,
        /// Buffers skipped by the `expect_dims` gate (rate-limits its log).
        gate_skips: u64,
        /// When the gate first held a buffer — after [`GATE_DEADLINE`] with no renegotiation the
        /// gate disarms and accepts what the producer serves (degraded dims beat a session wedged
        /// into the first-frame-timeout retry loop; the promised renegotiation normally lands
        /// within a frame or two).
        gate_since: Option<std::time::Instant>,
    }
    /// Consecutive tiled-import failures (worker alive, e.g. a per-buffer `EGL_BAD_MATCH`) before
@@ -1312,10 +1398,17 @@ mod pipewire {
                    value: pw::spa::pod::Value::Choice(pw::spa::pod::ChoiceValue::Int(
                        pw::spa::utils::Choice(
                            pw::spa::utils::ChoiceFlags::empty(),
                            // The max must cover the producer's offer or the Meta param silently
                            // fails to negotiate and NO buffer ever carries the meta region:
                            // Mutter offers a FIXED `SPA_POD_Int(CURSOR_META_SIZE(384, 384))`
                            // (meta-screen-cast-stream-src.c, GNOME 50) — a 256² max made the
                            // intersection empty, which cost the whole Linux cursor channel
                            // on-glass. 1024² is headroom, not an allocation: the negotiated
                            // region follows the producer's value.
                            pw::spa::utils::ChoiceEnum::Range {
                                default: meta_size(64, 64),
                                min: meta_size(1, 1),
-                                max: meta_size(256, 256),
+                                max: meta_size(1024, 1024),
                            },
                        ),
                    )),
@@ -1373,13 +1466,19 @@ mod pipewire {
            )
        };
        if id == 0 {
-            // Compositor reports no visible pointer (e.g. a game grabbed/hid it).
+            // SPA contract: id 0 = "no cursor information", NOT "cursor hidden". Mutter only
-            cursor.visible = false;
+            // REWRITES a buffer's meta region when the cursor changed, so recycled buffers
            // between damage frames carry a stale id-0 meta — treating that as hidden flickered
            // the cursor off between hovers (on-glass round 5). Keep the last-known state; a
            // pointer that really left/hid simply stops producing updates. (The M3 hidden hint
            // loses its Mutter signal — Windows has its own CURSOR_SUPPRESSED source.)
            return;
        }
        cursor.visible = true;
        cursor.x = pos_x - hot_x;
        cursor.y = pos_y - hot_y;
        cursor.hot_x = hot_x;
        cursor.hot_y = hot_y;
        if bmp_off == 0 {
            // Position-only update — keep the cached bitmap.
            return;
@@ -1405,8 +1504,9 @@ mod pipewire {
                (*bmp).offset as usize,
            )
        };
-        // Ignore empty or implausibly large bitmaps (we requested <= 256×256).
+        // Ignore empty or implausibly large bitmaps (the meta-size request covers <= 1024×1024;
-        if bw == 0 || bh == 0 || bw > 256 || bh > 256 {
+        // real cursors are ≤96px — the cursor channel downscales >120px for the wire anyway).
        if bw == 0 || bh == 0 || bw > 1024 || bh > 1024 {
            return;
        }
        let row = bw as usize * 4;
@@ -2006,6 +2106,9 @@ mod pipewire {
        streaming: Arc<AtomicBool>,
        broken: Arc<AtomicBool>,
        hdr_negotiated: Arc<AtomicBool>,
        // LIVE cursor publisher (see `PortalCapturer::cursor_live`): refreshed from every
        // dequeued buffer's cursor meta, frames or not.
        cursor_live: Arc<std::sync::Mutex<Option<pf_frame::CursorOverlay>>>,
        zerocopy: bool,
        // 4:4:4 session: tiled dmabufs take the worker's planar-YUV444 GPU convert.
        want_444: bool,
@@ -2017,6 +2120,9 @@ mod pipewire {
        // Encode-backend facts resolved by the facade (never re-derived here) — the one-way
        // capture→encode edge (plan §W6).
        policy: ZeroCopyPolicy,
        // See `spawn_pipewire`: the first negotiation is for a sacrificial mode; hold frames
        // until the producer renegotiates to `preferred`'s dims.
        expect_exact_dims: bool,
    ) -> Result<()> {
        crate::pwinit::ensure_init();
@@ -2201,6 +2307,14 @@ mod pipewire {
            linear_nv12_failed: false,
            dbg_log_n: 0,
            cursor: CursorState::default(),
            cursor_live,
            expect_dims: if expect_exact_dims {
                preferred.map(|(w, h, _)| (w, h))
            } else {
                None
            },
            gate_skips: 0,
            gate_since: None,
        };
        let stream = pw::stream::StreamBox::new(
@@ -2312,6 +2426,60 @@ mod pipewire {
                    newest = next;
                    drained += 1;
                }
                // Sacrificial-mode gate (kwin.rs `create`): until the producer renegotiates to the
                // expected dims, every buffer — frame AND cursor meta, whose positions are in the
                // doomed mode's space — belongs to the birth mode; consuming one would build the
                // pipeline at the wrong size. Self-disarms on the first matching negotiation, or
                // after `GATE_DEADLINE` without one — degraded dims beat wedging the session into
                // the first-frame-timeout retry loop when the promised renegotiation never comes.
                if let Some((ew, eh)) = ud.expect_dims {
                    /// The renegotiation normally lands within a frame or two of recording; well
                    /// past that, the producer is not going to deliver it (the on-glass case: the
                    /// real mode never actually applied) — stop starving the pipeline.
                    const GATE_DEADLINE: std::time::Duration = std::time::Duration::from_secs(3);
                    let sz = ud.info.size();
                    if sz.width == ew && sz.height == eh {
                        tracing::info!(
                            skipped = ud.gate_skips,
                            width = ew,
                            height = eh,
                            "producer renegotiated to the expected mode — frames flow"
                        );
                        ud.expect_dims = None;
                    } else if ud
                        .gate_since
                        .get_or_insert_with(std::time::Instant::now)
                        .elapsed()
                        > GATE_DEADLINE
                    {
                        tracing::warn!(
                            negotiated_w = sz.width,
                            negotiated_h = sz.height,
                            expected_w = ew,
                            expected_h = eh,
                            skipped = ud.gate_skips,
                            "producer never renegotiated to the expected mode — accepting its \
                             dims (session runs degraded rather than wedged)"
                        );
                        ud.expect_dims = None;
                    } else {
                        ud.gate_skips += 1;
                        if ud.gate_skips == 1 || ud.gate_skips.is_power_of_two() {
                            tracing::info!(
                                negotiated_w = sz.width,
                                negotiated_h = sz.height,
                                expected_w = ew,
                                expected_h = eh,
                                n = ud.gate_skips,
                                "holding frames until the producer renegotiates to the expected mode"
                            );
                        }
                        // SAFETY: `newest` was dequeued from this stream and not yet requeued;
                        // requeued exactly once here, then never touched (mirrors the null path).
                        unsafe { stream.queue_raw_buffer(newest) };
                        return;
                    }
                }
                // PipeWire dispatches from a C trampoline with no catch_unwind; a panic crossing that FFI
                // boundary would abort the whole host. Contain the inspect/consume work — the only Rust
                // code here that can panic — and requeue `newest` unconditionally after it.
@@ -2325,6 +2493,19 @@ mod pipewire {
                    // pointer-only movements as metadata-only "corrupted" buffers we drop for their
                    // frame, but their cursor meta is fresh and must still move our overlay.
                    update_cursor_meta(&mut ud.cursor, spa_buf);
                    // Publish the LIVE overlay (frames or not) so the encode loop's forwarder
                    // tracks pointer-only motion on a static desktop — the frame-attached overlay
                    // alone stales between damage frames. ONLY when we actually have one: a
                    // gamescope node carries no `SPA_META_Cursor`, so `overlay()` is always `None`
                    // here, and writing that would clobber — at frame rate — the `Some` the
                    // attached XFixes source publishes into this SAME slot, strobing the
                    // composited pointer on/off. Portal cursors are `None` only before the first
                    // bitmap (nothing to drop), and a HIDDEN pointer is still `Some(visible:false)`.
                    if let Some(overlay) = ud.cursor.overlay() {
                        if let Ok(mut slot) = ud.cursor_live.lock() {
                            *slot = Some(overlay);
                        }
                    }
                    // Inspect the newest buffer's header + first chunk for the diagnostic and the
                    // CORRUPTED skip. SPA_META_Header is optional — `hdr` may be null.
@@ -0,0 +1,366 @@
 //! XFixes cursor source for the gamescope capture path (remote-desktop-sweep Phase C).
 //!
 //! gamescope draws the pointer on a DRM hardware-cursor plane and its `paint_pipewire()`
 //! deliberately excludes the cursor from the frame it feeds its built-in PipeWire node — so
 //! `SPA_META_Cursor` never arrives and the ordinary [`cursor_live`](super::PortalCapturer) slot
 //! stays empty (a KWin/GNOME session gets its cursor from that meta; gamescope can't embed one
 //! either, its `set_hw_cursor` is inert). We instead read the pointer from gamescope's nested
 //! Xwayland via X11 — the trick Sunshine uses — and publish a [`CursorOverlay`] into that same
 //! slot, so the encoder blend composites the pointer into the video exactly like the portal path.
 //!
 //! **Multiple Xwaylands.** gamescope runs one Xwayland per `--xwayland-count` (Steam Gaming Mode
 //! uses 2: one for Big Picture, one for the game). The pointer lives on whichever is FOCUSED — an
 //! inactive display's pointer is frozen. So the source connects to ALL of them and each tick
 //! follows the one whose pointer actually moved (gamescope routes input to the focused surface, so
 //! exactly one moves at a time). It reads that display's shape too, since each Xwayland has its own
 //! current cursor. This is why a single-display read froze the pointer the moment a game on the
 //! OTHER Xwayland took focus.
 //!
 //! Two X sources per display, split by cost (Sunshine's split):
 //! * **Position** — core `QueryPointer` on the root, polled fast. Cheap (a few-byte reply, no
 //!   bitmap), so it can out-pace the stream fps and keep the composited pointer smooth. It also
 //!   doubles as the focus signal (the display whose pointer moves is the active one).
 //! * **Shape / hotspot / visibility** — `XFixesGetCursorImage`, refreshed only after an XFixes
 //!   `CursorNotify` (a real cursor change). A game hiding the pointer IS a cursor change → the
 //!   image comes back fully transparent → `visible: false`, which the encode loop strips before
 //!   any blend path draws it (so a grabbed pointer shows nothing, matching native gamescope).
 use std::sync::{
    atomic::{AtomicBool, Ordering},
    Arc, Mutex,
 };
 use std::time::Duration;
 use pf_frame::CursorOverlay;
 use x11rb::connection::Connection;
 use x11rb::errors::ReplyError;
 use x11rb::protocol::xfixes::{self, ConnectionExt as _, GetCursorImageReply};
 use x11rb::protocol::xproto::{ConnectionExt as _, QueryPointerReply, Window};
 use x11rb::rust_connection::RustConnection;
 /// Serializes the brief `XAUTHORITY` env swap around a connect (the var is process-global). Only
 /// ever contended if two gamescope sessions start at once — rare, and the swap is microseconds.
 static XAUTH_LOCK: Mutex<()> = Mutex::new(());
 /// Position out-paces the stream fps (`POLL`); shape rides `CursorNotify` events drained each tick.
 /// 4 ms ≈ 250 Hz matches the Windows GDI poller — the polled position IS the composited position
 /// and must out-run a 240 fps session or the pointer stutters.
 const POLL: Duration = Duration::from_millis(4);
 /// A running XFixes cursor reader. Dropping it stops the worker thread and joins it, releasing the
 /// X connections — so it lives exactly as long as the capturer that owns it.
 pub(super) struct XFixesCursorSource {
    stop: Arc<AtomicBool>,
    join: Option<std::thread::JoinHandle<()>>,
 }
 impl XFixesCursorSource {
    /// Connect to every gamescope nested Xwayland in `targets` (`(DISPLAY, XAUTHORITY)`) and start
    /// publishing cursor overlays into `slot`, following the focused display's pointer. Returns
    /// `None` — and logs — if NONE can be used (no X connection / no XFixes), so the caller
    /// degrades to no gamescope cursor (today's behaviour) instead of failing the session.
    pub(super) fn spawn(
        targets: Vec<(String, Option<String>)>,
        slot: Arc<Mutex<Option<CursorOverlay>>>,
    ) -> Option<Self> {
        // Connect on the caller's thread so failures degrade cleanly and the displays are validated
        // before we commit a thread.
        let mut displays = Vec::new();
        for (dpy, xauth) in targets {
            match connect(&dpy, xauth.as_deref()) {
                Ok((conn, root)) => displays.push(XDisplay::new(dpy, conn, root)),
                Err(e) => tracing::warn!(
                    dpy = %dpy,
                    error = %e,
                    "gamescope cursor: skipping a nested Xwayland we can't use"
                ),
            }
        }
        if displays.is_empty() {
            tracing::warn!(
                "gamescope cursor: no usable nested Xwayland — no in-video pointer this session \
                 (falls back to today's cursorless gamescope stream)"
            );
            return None;
        }
        let names: Vec<&str> = displays.iter().map(|d| d.name.as_str()).collect();
        tracing::info!(
            displays = ?names,
            "gamescope cursor: XFixes source live — following the focused Xwayland's pointer"
        );
        let stop = Arc::new(AtomicBool::new(false));
        let stop_worker = Arc::clone(&stop);
        let join = std::thread::Builder::new()
            .name("pf-gs-cursor".into())
            .spawn(move || run(displays, slot, stop_worker))
            .ok()?;
        Some(XFixesCursorSource {
            stop,
            join: Some(join),
        })
    }
 }
 impl Drop for XFixesCursorSource {
    fn drop(&mut self) {
        self.stop.store(true, Ordering::Relaxed);
        if let Some(j) = self.join.take() {
            let _ = j.join();
        }
    }
 }
 /// Open the X connection, negotiate XFixes, and select cursor-change events — returning the
 /// connection + root window. `RustConnection` reads `XAUTHORITY` from the env at connect time only,
 /// so set it under the lock (the host isn't a gamescope child), connect, then restore.
 fn connect(dpy: &str, xauthority: Option<&str>) -> Result<(RustConnection, Window), String> {
    let (conn, screen_num) = {
        let _g = XAUTH_LOCK.lock().unwrap_or_else(|e| e.into_inner());
        let prev = std::env::var_os("XAUTHORITY");
        if let Some(x) = xauthority {
            std::env::set_var("XAUTHORITY", x);
        }
        let out = RustConnection::connect(Some(dpy));
        match (&prev, xauthority) {
            (Some(p), _) => std::env::set_var("XAUTHORITY", p),
            (None, Some(_)) => std::env::remove_var("XAUTHORITY"),
            (None, None) => {}
        }
        out.map_err(|e| format!("connect: {e}"))?
    };
    // XFixes ≥ 1 gives GetCursorImage / SelectCursorInput; ask for a modern minor, take what we get.
    conn.xfixes_query_version(5, 0)
        .map_err(ReplyError::from)
        .and_then(|c| c.reply())
        .map_err(|e| format!("XFixes unavailable: {e}"))?;
    let root = conn
        .setup()
        .roots
        .get(screen_num)
        .ok_or_else(|| format!("no X screen {screen_num}"))?
        .root;
    // Wake the worker's event drain whenever the cursor shape changes (incl. hide/show).
    conn.xfixes_select_cursor_input(root, xfixes::CursorNotifyMask::DISPLAY_CURSOR)
        .map_err(ReplyError::from)
        .and_then(|c| c.check())
        .map_err(|e| format!("SelectCursorInput: {e}"))?;
    let _ = conn.flush();
    Ok((conn, root))
 }
 /// One gamescope Xwayland the source tracks.
 struct XDisplay {
    name: String,
    conn: RustConnection,
    root: Window,
    /// Last polled pointer position — a change since the previous tick marks this display FOCUSED.
    last_pos: Option<(i32, i32)>,
    /// Cached cursor shape, refreshed only after this display's XFixes `CursorNotify`.
    shape: Shape,
    /// A `CursorNotify` (or first read) is pending — fetch the shape when this display is active.
    need_shape: bool,
    /// The X connection died (game/Xwayland exited) — skip it.
    dead: bool,
 }
 impl XDisplay {
    fn new(name: String, conn: RustConnection, root: Window) -> Self {
        XDisplay {
            name,
            conn,
            root,
            last_pos: None,
            shape: Shape::default(),
            need_shape: true,
            dead: false,
        }
    }
 }
 /// Cached cursor shape for one display.
 #[derive(Default)]
 struct Shape {
    /// Straight-alpha RGBA (`w*h*4`, bytes R,G,B,A); empty before the first image arrives.
    rgba: Arc<Vec<u8>>,
    w: u32,
    h: u32,
    hot_x: u32,
    hot_y: u32,
    /// XFixes' own per-display cursor serial — bumps on every shape change.
    serial: u64,
    /// A hidden pointer arrives as an all-transparent image; kept so a position-only tick preserves
    /// the last known visibility.
    visible: bool,
 }
 fn run(
    mut displays: Vec<XDisplay>,
    slot: Arc<Mutex<Option<CursorOverlay>>>,
    stop: Arc<AtomicBool>,
 ) {
    let mut active = 0usize;
    // The overlay serial must bump whenever the DRAWN cursor changes — either the active display's
    // shape OR which display is active (per-display XFixes serials aren't comparable across
    // displays, so switching could reuse a number and the encoder would keep the old texture).
    let mut out_serial = 0u64;
    let mut last_key = (usize::MAX, u64::MAX);
    let mut warned_image = false;
    while !stop.load(Ordering::Relaxed) {
        // 1) Poll every display's pointer; note which moved since last tick (the focus signal).
        let mut active_moved = false;
        let mut other_moved: Option<usize> = None;
        for (i, d) in displays.iter_mut().enumerate() {
            if d.dead {
                continue;
            }
            // Drain pending events; the only ones selected are CursorNotify, so ANY event means
            // "re-read this display's shape". poll_for_event never blocks.
            loop {
                match d.conn.poll_for_event() {
                    Ok(Some(_)) => d.need_shape = true,
                    Ok(None) => break,
                    Err(_) => {
                        d.dead = true;
                        break;
                    }
                }
            }
            match fetch_pointer(&d.conn, d.root) {
                Ok(p) if p.same_screen => {
                    let pos = (i32::from(p.root_x), i32::from(p.root_y));
                    let moved = d.last_pos.is_some_and(|lp| lp != pos);
                    d.last_pos = Some(pos);
                    if moved {
                        if i == active {
                            active_moved = true;
                        } else if other_moved.is_none() {
                            other_moved = Some(i);
                        }
                    }
                }
                Ok(_) => {} // pointer on another screen — keep the last position.
                Err(_) => d.dead = true,
            }
        }
        // 2) Switch focus: sticky to the active display while it moves (no flapping); otherwise
        //    follow another display that moved. If the active one died, fall to any live display.
        if !active_moved {
            if let Some(j) = other_moved {
                active = j;
            }
        }
        if displays.get(active).is_none_or(|d| d.dead) {
            match displays.iter().position(|d| !d.dead) {
                Some(k) => active = k,
                None => {
                    std::thread::sleep(POLL); // all connections dead — idle until Drop.
                    continue;
                }
            }
        }
        // 3) Fetch the active display's shape if a CursorNotify (or a focus switch) left it stale.
        if displays[active].need_shape {
            match fetch_cursor_image(&displays[active].conn) {
                Ok(img) => {
                    update_shape(&mut displays[active].shape, &img);
                    displays[active].need_shape = false;
                }
                Err(e) => {
                    if !warned_image {
                        warned_image = true;
                        tracing::warn!(error = %e, "gamescope cursor: GetCursorImage failed — retrying");
                    }
                }
            }
        }
        // 4) Publish the ACTIVE display's pointer + shape (or clear the slot when it has no cursor
        //    of its own — so a focus switch never leaves the other display's stale pointer showing).
        let d = &displays[active];
        let overlay = match (d.last_pos, d.shape.rgba.is_empty()) {
            (Some((px, py)), false) => {
                let key = (active, d.shape.serial);
                if key != last_key {
                    out_serial += 1;
                    last_key = key;
                }
                Some(CursorOverlay {
                    // Top-left = pointer position − hotspot (the overlay contract).
                    x: px - d.shape.hot_x as i32,
                    y: py - d.shape.hot_y as i32,
                    w: d.shape.w,
                    h: d.shape.h,
                    rgba: Arc::clone(&d.shape.rgba),
                    serial: out_serial,
                    hot_x: d.shape.hot_x,
                    hot_y: d.shape.hot_y,
                    visible: d.shape.visible,
                })
            }
            _ => None,
        };
        if let Ok(mut s) = slot.lock() {
            *s = overlay;
        }
        std::thread::sleep(POLL);
    }
 }
 /// Update `shape` from a fresh `GetCursorImage` reply. A hidden pointer (all-transparent) keeps the
 /// last bitmap (instant re-show) but flips visibility; the serial still bumps so the change shows.
 fn update_shape(shape: &mut Shape, img: &GetCursorImageReply) {
    let visible =
        img.width > 0 && img.height > 0 && img.cursor_image.iter().any(|&p| (p >> 24) & 0xff != 0);
    if visible {
        shape.rgba = Arc::new(argb_premul_to_straight_rgba(&img.cursor_image));
        shape.w = u32::from(img.width);
        shape.h = u32::from(img.height);
        shape.hot_x = u32::from(img.xhot);
        shape.hot_y = u32::from(img.yhot);
    }
    shape.visible = visible;
    shape.serial = u64::from(img.cursor_serial);
 }
 /// One request+reply — x11rb splits errors (the request is `ConnectionError`, `reply()` is
 /// `ReplyError` which is `From<ConnectionError>`), so the request `?` converts into the reply error.
 fn fetch_cursor_image(conn: &RustConnection) -> Result<GetCursorImageReply, ReplyError> {
    conn.xfixes_get_cursor_image()?.reply()
 }
 fn fetch_pointer(conn: &RustConnection, root: Window) -> Result<QueryPointerReply, ReplyError> {
    conn.query_pointer(root)?.reply()
 }
 /// XFixes cursor pixels are packed `0xAARRGGBB` with **premultiplied** alpha (the Xrender / Xcursor
 /// convention). The overlay + both blend paths want **straight** alpha RGBA (R,G,B,A bytes), like
 /// the `SPA_META_Cursor` path — so un-premultiply here. (If on-glass shows over-bright fringes the
 /// source wasn't premultiplied after all; drop the divide.)
 fn argb_premul_to_straight_rgba(argb: &[u32]) -> Vec<u8> {
    let mut out = Vec::with_capacity(argb.len() * 4);
    for &px in argb {
        let a = (px >> 24) & 0xff;
        let r = (px >> 16) & 0xff;
        let g = (px >> 8) & 0xff;
        let b = px & 0xff;
        let (r, g, b) = match a {
            0 => (0, 0, 0),
            255 => (r, g, b),
            a => (
                ((r * 255 + a / 2) / a).min(255),
                ((g * 255 + a / 2) / a).min(255),
                ((b * 255 + a / 2) / a).min(255),
            ),
        };
        out.extend_from_slice(&[r as u8, g as u8, b as u8, a as u8]);
    }
    out
 }
@@ -161,7 +161,7 @@ pub fn install_gpu_pref_hook() {
    });
 }
-unsafe fn compile_shader(src: &str, entry: PCSTR, target: PCSTR) -> Result<Vec<u8>> {
+pub(crate) unsafe fn compile_shader(src: &str, entry: PCSTR, target: PCSTR) -> Result<Vec<u8>> {
    let mut blob: Option<ID3DBlob> = None;
    let mut errs: Option<ID3DBlob> = None;
    let r = D3DCompile(
@@ -194,7 +194,7 @@ unsafe fn compile_shader(src: &str, entry: PCSTR, target: PCSTR) -> Result<Vec<u
 }
 /// Fullscreen-triangle vertex shader for the HDR conversion pass (3 verts, no input layout).
-const HDR_VS: &str = r"
+pub(crate) const HDR_VS: &str = r"
 struct VOut { float4 pos : SV_POSITION; float2 uv : TEXCOORD0; };
 VOut main(uint vid : SV_VertexID) {
    float2 uv = float2((vid << 1) & 2, vid & 2);
@@ -365,6 +365,12 @@ pub unsafe fn verify_is_wudfhost(process: HANDLE, wudf_pid: u32, what: &str) ->
 #[path = "idd_push/channel.rs"]
 mod channel;
 #[path = "idd_push/cursor.rs"]
 mod cursor;
 #[path = "idd_push/cursor_blend.rs"]
 mod cursor_blend;
 #[path = "idd_push/cursor_poll.rs"]
 mod cursor_poll;
 #[path = "idd_push/descriptor.rs"]
 mod descriptor;
 #[path = "idd_push/stall.rs"]
@@ -386,6 +392,64 @@ pub struct IddPushCapturer {
    /// The sealed channel's handle-duplication broker (WUDFHost process + control device); used at open
    /// and again on every ring recreate to deliver fresh duplicates.
    broker: ChannelBroker,
    /// The v5 hardware-cursor channel's host end (`Some` = delivered; the driver declared the
    /// hardware cursor and seqlock-publishes into it). Survives ring recreates — the section is
    /// independent of the frame ring's generation. With the channel delivered, the driver's
    /// hardware cursor keeps DWM from compositing ANY cursor into the frame; the SHAPE now comes
    /// from [`cursor_poll::CursorPoller`] (the IddCx query is alpha-only — see cursor_poll.rs),
    /// and this shm read is the fallback if that poller dies.
    cursor_shared: Option<cursor::CursorShared>,
    /// The GDI cursor-shape poller (design §8): the overlay source while alive. `Some` when
    /// `cursor_shared` is (both ride the negotiated cursor channel + successful delivery) — or
    /// when `composite_forced` (no channel, but the target's sticky declare needs a blend source).
    cursor_poll: Option<cursor_poll::CursorPoller>,
    /// Retained delivery sender (`IOCTL_SET_CURSOR_CHANNEL`) for RE-delivery: a driver-side
    /// monitor re-arrival (match-window re-arrival resize, a sibling session recreating the
    /// shared slot) destroys the driver's cursor worker — the section here survives, so the
    /// channel is re-delivered on ring recreates.
    cursor_sender: Option<crate::CursorChannelSender>,
    /// The cursor-render flip sender (`IOCTL_SET_CURSOR_FORWARD`) — the secure-desktop guard's
    /// actuator. UAC/Winlogon render only through the OS's software-cursor path (its default on
    /// every mode commit); with our hardware cursor declared (and re-declared on every
    /// swap-chain assign) that path never comes back, and the secure desktop never presents —
    /// the stream freezes on the last normal-desktop frame for the whole UAC/lock interaction.
    /// [`Self::poll_secure_desktop`] flips the declare off/on at the secure-desktop edges.
    cursor_forward: Option<crate::CursorForwardSender>,
    /// The secure-desktop guard's edge state: `true` = the poller reports a secure input
    /// desktop and the declare is currently stood down.
    secure_active: bool,
    /// The CAPTURE mouse model is active — the HOST composites the pointer into the frame
    /// (see cursor_blend.rs for why DWM cannot: a declared IddCx hardware cursor is forever).
    composite_cursor: bool,
    /// This session never negotiated the cursor channel but its target carries an IRREVOCABLE
    /// hardware-cursor declare from an earlier session (`WinCaptureTarget::cursor_excluded`,
    /// §8.6): DWM delivers pointer-free frames and no client draws the cursor, so the ONLY path
    /// to a visible pointer is compositing here. Pins `composite_cursor` on — nothing may turn
    /// it off (there is no channel to hand the pointer to).
    composite_forced: bool,
    /// The cursor-quad blend pass (lazy; per capture device). `None` after a build failure —
    /// composite mode then degrades to pointer-less frames (warned once).
    cursor_blend: Option<cursor_blend::CursorBlendPass>,
    cursor_blend_failed: bool,
    /// The frame-sized blend scratch (slot copy + cursor quad): texture + SRV + (w, h, fmt)
    /// it was built for — rebuilt when the ring geometry changes.
    blend_scratch: Option<(
        ID3D11Texture2D,
        ID3D11ShaderResourceView,
        u32,
        u32,
        DXGI_FORMAT,
    )>,
    /// The (serial, x, y, visible) of the LAST blended pointer — the composite-regen change
    /// key: pointer-only motion produces no driver publish (the declared hardware cursor
    /// doesn't dirty frames), so `try_consume` regenerates from the last slot when this moves.
    last_blend_key: Option<(u64, i32, i32, bool)>,
    /// The ring slot of the last FRESH publish — the regen source.
    last_slot: Option<usize>,
    /// The target's SDR-white scale (vs 80 nits) for HDR cursor compositing — refreshed on
    /// each blend-scratch rebuild (first use + ring geometry changes). 2.5 ≈ the Windows
    /// SDR-brightness default; without it the composited cursor renders visibly dark on HDR.
    sdr_white_scale: f32,
    width: u32,
    height: u32,
    slots: Vec<HostSlot>,
@@ -611,11 +675,22 @@ impl IddPushCapturer {
        pyrowave: bool,
        keepalive: Box<dyn Send>,
        sender: crate::FrameChannelSender,
        cursor_sender: Option<crate::CursorChannelSender>,
        cursor_forward: Option<crate::CursorForwardSender>,
    ) -> std::result::Result<Self, (anyhow::Error, Box<dyn Send>)> {
        // The stall-attribution listener (idempotent): started with the first IDD-push capturer so
        // the stall log can correlate DWM holes with OS display events for the session's lifetime.
        pf_win_display::display_events::spawn_once();
-        match Self::open_inner(target, preferred, client_10bit, want_444, pyrowave, sender) {
+        match Self::open_inner(
            target,
            preferred,
            client_10bit,
            want_444,
            pyrowave,
            sender,
            cursor_sender,
            cursor_forward,
        ) {
            Ok(mut me) => {
                me._keepalive = keepalive;
                Ok(me)
@@ -632,6 +707,8 @@ impl IddPushCapturer {
        want_444: bool,
        pyrowave: bool,
        sender: crate::FrameChannelSender,
        cursor_sender: Option<crate::CursorChannelSender>,
        cursor_forward: Option<crate::CursorForwardSender>,
    ) -> 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
@@ -654,6 +731,8 @@ impl IddPushCapturer {
            pyrowave,
            luid,
            sender.clone(),
            cursor_sender.clone(),
            cursor_forward.clone(),
        ) {
            Ok(me) => Ok(me),
            Err(e) => {
@@ -687,6 +766,8 @@ impl IddPushCapturer {
                    pyrowave,
                    drv,
                    sender,
                    cursor_sender,
                    cursor_forward,
                )
                .context("IDD-push rebind to the driver's reported render adapter")
            }
@@ -702,6 +783,8 @@ impl IddPushCapturer {
        pyrowave: bool,
        luid: LUID,
        sender: crate::FrameChannelSender,
        cursor_sender: Option<crate::CursorChannelSender>,
        cursor_forward: Option<crate::CursorForwardSender>,
    ) -> Result<Self> {
        let (pw, ph, _hz) = preferred
            .context("IDD push needs the negotiated mode (WxH) to size the shared ring")?;
@@ -935,6 +1018,61 @@ impl IddPushCapturer {
                )
                .context("deliver IDD-push frame channel to the driver")?;
            // v5 hardware-cursor channel (M2c): create + deliver the CursorShm section. Failure
            // is NON-fatal — the driver never declares the hardware cursor without this delivery,
            // so the session degrades to today's composited pointer (and the forwarder simply
            // never sees a live overlay).
            let cursor_shared = cursor_sender.as_ref().and_then(|send_cursor| {
                match cursor::CursorShared::create(target.target_id) {
                    Ok(cs) => {
                        // Deliver via the shared helper (also used for RE-delivery after a
                        // driver-side monitor re-arrival destroyed the worker).
                        deliver_cursor_channel(&broker, target.target_id, &cs, send_cursor)
                            .then_some(cs)
                    }
                    Err(e) => {
                        tracing::warn!(
                            "cursor section creation failed (composited cursor stays): {e:#}"
                        );
                        None
                    }
                }
            });
            // No channel this session, but the target's sticky declare (an EARLIER session's —
            // irrevocable, §8.6) keeps DWM's frames pointer-free with no client drawing either:
            // the only visible pointer is the one composited here, so force composite mode on.
            let composite_forced = target.cursor_excluded && cursor_sender.is_none();
            if composite_forced {
                tracing::info!(
                    target_id = target.target_id,
                    "target carries an irrevocable hardware-cursor declare from an earlier \
                     desktop-mode session and this session has no cursor channel — the host \
                     composites the pointer into frames (forced, for the session's life)"
                );
            }
            // The GDI shape poller rides the SAME gate as the delivered channel: with the driver's
            // hardware cursor keeping the frame cursor-free, the poller supplies the full-fidelity
            // shape (masked/monochrome included — the IddCx query can't; see cursor_poll.rs).
            // Forced-composite sessions need it too — it is their only shape/position source.
            let cursor_poll = (cursor_shared.is_some() || composite_forced).then(|| {
                // Safety of the CCD call: read-only QueryDisplayConfig over owned locals (same
                // call CursorShared::create makes) — already inside open_on's unsafe region.
                let rect = pf_win_display::win_display::source_desktop_rect(target.target_id)
                    .unwrap_or((0, 0, i32::MAX, i32::MAX));
                cursor_poll::CursorPoller::spawn(target.target_id, rect)
            });
            // Heal the driver's persisted cursor-forward state: a session that died on the
            // secure desktop (client drops at the lock screen — the common case) leaves the
            // per-target desired state `false`, and the NEXT session's channel delivery would
            // adopt UNdeclared (the exact cross-session composite trap of §8.6). A fresh
            // session always starts declared; the secure-desktop guard re-disables if the
            // secure desktop is (still) up, via its first `poll_secure_desktop` edge.
            if let (Some(_), Some(fwd)) = (cursor_shared.as_ref(), cursor_forward.as_ref()) {
                if let Err(e) = fwd(true) {
                    tracing::debug!("cursor-forward reset at open failed (pre-v6 driver?): {e:#}");
                }
            }
            tracing::info!(
                target_id = target.target_id,
                wudf_pid = target.wudf_pid,
@@ -993,6 +1131,19 @@ impl IddPushCapturer {
                last_seq: 0,
                last_present: None,
                status_logged: false,
                cursor_shared,
                cursor_poll,
                cursor_sender,
                cursor_forward,
                secure_active: false,
                composite_cursor: composite_forced,
                composite_forced,
                cursor_blend: None,
                cursor_blend_failed: false,
                blend_scratch: None,
                last_blend_key: None,
                last_slot: None,
                sdr_white_scale: 1.0,
                // Held from BEFORE the first-frame gate (the display must not idle off while we
                // wait for the first compose) until the capturer drops with the session.
                _display_wake: pf_frame::session_tuning::DisplayWakeRequest::new(),
@@ -1337,6 +1488,14 @@ impl IddPushCapturer {
                "IDD push: frame-channel re-delivery failed after ring recreate"
            );
        }
        // Ring recreates ride display churn that can also have re-arrived the MONITOR driver-side
        // (destroying its cursor worker with it) — re-deliver the surviving cursor section so the
        // hardware-cursor declaration follows the CURRENT monitor generation.
        if let (Some(cs), Some(send)) = (self.cursor_shared.as_ref(), self.cursor_sender.as_ref()) {
            let _ = deliver_cursor_channel(&self.broker, self.target_id, cs, send);
        }
        self.blend_scratch = None; // ring geometry/format changed — rebuild at next blend
        self.last_slot = None; // old-ring slot indices are meaningless now
        self.last_seq = 0;
        self.out_ring.clear(); // the output format changed → rebuild lazily at the new format
        self.video_conv = None; // converters are sized + HDR-specific → rebuild at the new mode
@@ -1622,8 +1781,201 @@ impl IddPushCapturer {
        Ok(Some((self.pyro_fence_handle, value)))
    }
    /// The (serial, x, y, visible) of the CURRENT polled cursor — the composite-regen change
    /// key. `None` while the poller has no shape yet (or isn't running).
    fn cursor_blend_key(&self) -> Option<(u64, i32, i32, bool)> {
        self.cursor_poll
            .as_ref()
            .and_then(|p| p.read())
            .map(|o| (o.serial, o.x, o.y, o.visible))
    }
    /// Composite the pointer for this convert: ensure the frame-sized blend scratch, copy the
    /// slot into it, and alpha-blend the GDI poller's shape at its polled position. Returns the
    /// scratch (texture + SRV) the conversion should read INSTEAD of the slot; `None` degrades
    /// to the pointer-less slot (scratch/pass creation failed — warned once). A hidden pointer
    /// blends nothing (the plain copy is the correct frame).
    ///
    /// # Safety
    /// D3D11 calls on the owning capture/encode thread's device + immediate context, called
    /// while holding the slot's keyed mutex (the copy reads the slot).
    unsafe fn prepare_blend_scratch(
        &mut self,
        slot_tex: &ID3D11Texture2D,
    ) -> Option<(ID3D11Texture2D, ID3D11ShaderResourceView)> {
        let fmt = self.ring_format();
        // (Re)build the scratch at the current ring geometry.
        let stale = self
            .blend_scratch
            .as_ref()
            .is_none_or(|(_, _, w, h, f)| (*w, *h, *f) != (self.width, self.height, fmt));
        if stale {
            self.blend_scratch = None;
            let desc = D3D11_TEXTURE2D_DESC {
                Width: self.width,
                Height: self.height,
                MipLevels: 1,
                ArraySize: 1,
                Format: fmt,
                SampleDesc: DXGI_SAMPLE_DESC {
                    Count: 1,
                    Quality: 0,
                },
                Usage: D3D11_USAGE_DEFAULT,
                BindFlags: (D3D11_BIND_RENDER_TARGET.0 | D3D11_BIND_SHADER_RESOURCE.0) as u32,
                ..Default::default()
            };
            let mut tex: Option<ID3D11Texture2D> = None;
            let built = self
                .device
                .CreateTexture2D(&desc, None, Some(&mut tex))
                .ok()
                .and(tex)
                .and_then(|t| {
                    let mut srv: Option<ID3D11ShaderResourceView> = None;
                    self.device
                        .CreateShaderResourceView(&t, None, Some(&mut srv))
                        .ok()
                        .and(srv)
                        .map(|v| (t, v))
                });
            match built {
                Some((t, v)) => {
                    self.blend_scratch = Some((t, v, self.width, self.height, fmt));
                    if self.display_hdr {
                        // Where DWM places SDR white on this HDR desktop — the composited
                        // cursor must match or it reads dark (~2.5x at the Windows default).
                        // Queried only here: scratch rebuilds are rare, and the CCD query
                        // contends on the display-config lock, which must stay OFF the
                        // per-frame path.
                        // Safety: read-only CCD query over owned locals (within unsafe fn).
                        let queried =
                            pf_win_display::win_display::sdr_white_level_scale(self.target_id);
                        self.sdr_white_scale = queried.unwrap_or(self.sdr_white_scale);
                        tracing::info!(
                            target_id = self.target_id,
                            queried = ?queried,
                            applied = self.sdr_white_scale,
                            "cursor composite: HDR SDR-white scale (1.0 = 80 nits; None = \
                             query failed — keeping the prior value)"
                        );
                    }
                }
                None => {
                    if !self.cursor_blend_failed {
                        self.cursor_blend_failed = true;
                        tracing::warn!(
                            "cursor blend scratch creation failed — capture-model frames stay \
                             pointer-less this session"
                        );
                    }
                    return None;
                }
            }
        }
        let (tex, srv, ..) = self.blend_scratch.as_ref().expect("just ensured");
        let (tex, srv) = (tex.clone(), srv.clone());
        self.context.CopyResource(&tex, slot_tex);
        // Blend the pointer (visible shapes only; hidden = the copy alone is the frame).
        let overlay = self.cursor_poll.as_ref().and_then(|p| p.read());
        self.last_blend_key = overlay.as_ref().map(|o| (o.serial, o.x, o.y, o.visible));
        if let Some(ov) = overlay.filter(|o| o.visible) {
            if self.cursor_blend.is_none() && !self.cursor_blend_failed {
                match cursor_blend::CursorBlendPass::new(&self.device) {
                    Ok(p) => self.cursor_blend = Some(p),
                    Err(e) => {
                        self.cursor_blend_failed = true;
                        tracing::warn!(
                            "cursor blend pass build failed — capture-model frames stay \
                             pointer-less this session: {e:#}"
                        );
                    }
                }
            }
            if let Some(pass) = self.cursor_blend.as_mut() {
                // FP16 ring = scRGB linear composition (HDR): linearize the sRGB shape and
                // scale it to the target's SDR white so it matches the desktop around it.
                let scale = if self.display_hdr {
                    self.sdr_white_scale
                } else {
                    0.0
                };
                if let Err(e) = pass.blend(&self.device, &self.context, &tex, &ov, scale) {
                    if !self.cursor_blend_failed {
                        self.cursor_blend_failed = true;
                        tracing::warn!("cursor blend draw failed — pointer-less frames: {e:#}");
                    }
                }
            }
        }
        Some((tex, srv))
    }
    /// The secure-desktop guard (the 0.18.0 UAC/Winlogon regression). UAC consent and Winlogon
    /// live on the SECURE desktop, which the OS renders through the software-cursor path — its
    /// per-mode-commit default. With this session's IddCx hardware cursor declared (and
    /// re-declared by the driver on every swap-chain assign), that path never materialises, the
    /// secure desktop never presents into our swap-chain, and the stream freezes on the last
    /// normal-desktop frame for the entire UAC/lock interaction. On the poller's secure edge:
    /// stand the declare down (`SET_CURSOR_FORWARD` off — the driver stops its per-assign
    /// re-declare — plus the host facade's forced same-mode re-commit that actualises the
    /// software cursor); on dismissal, re-declare. Runs on the capture/encode thread every tick
    /// (it must keep running while frames are stalled — that is exactly the state it exits).
    fn poll_secure_desktop(&mut self) {
        let Some(fwd) = self.cursor_forward.as_ref() else {
            return;
        };
        // Sessions with a declare possibly in play: the channel session that declared it, and
        // the forced-composite session whose (reused) driver monitor may still run an earlier
        // session's cursor worker. A plain session on a clean target has no poller — no guard.
        if self.cursor_shared.is_none() && !self.composite_forced {
            return;
        }
        let secure = self
            .cursor_poll
            .as_ref()
            .is_some_and(|p| p.secure_desktop());
        if secure == self.secure_active {
            return;
        }
        self.secure_active = secure;
        if secure {
            tracing::info!(
                target_id = self.target_id,
                "secure desktop (UAC/Winlogon) active — standing the IddCx hardware-cursor \
                 declare down so the OS software-cursor path can render it"
            );
            if let Err(e) = fwd(false) {
                tracing::warn!(
                    "secure-desktop cursor-forward stand-down failed (secure content may stay \
                     invisible this session): {e:#}"
                );
            }
        } else {
            tracing::info!(
                target_id = self.target_id,
                "secure desktop dismissed — restoring the cursor render model"
            );
            // Re-declare only for the session that RUNS the cursor channel; a forced-composite
            // session never wanted the declare (leaving the driver's desired state off also
            // stops a reused worker's per-assign re-declares for good — the next channel
            // session's open-time reset re-arms it).
            if self.cursor_shared.is_some() {
                if let Err(e) = fwd(true) {
                    tracing::warn!(
                        "secure-desktop cursor-forward re-enable failed (client-drawn cursor \
                         may double with a composited one): {e:#}"
                    );
                }
            }
        }
    }
    fn try_consume(&mut self) -> Result<Option<CapturedFrame>> {
        self.log_driver_status_once();
        // The secure-desktop guard first: while UAC/Winlogon is up there may be NO fresh frames
        // at all — this edge is what brings them back.
        self.poll_secure_desktop();
        // Follow the display: a "Use HDR" flip recreates the ring at the matching format.
        self.poll_display_hdr();
        // Recover-or-drop (GB1): if a descriptor change triggered a recreate but no fresh frame has resumed
@@ -1682,10 +2034,26 @@ impl IddPushCapturer {
            return Ok(None);
        }
        let seq = u64::from(tok.seq);
-        let slot = tok.slot as usize;
+        let mut slot = tok.slot as usize;
-        if seq == self.last_seq || slot >= self.slots.len() {
+        let fresh = seq != self.last_seq && slot < self.slots.len();
        let mut regen = false;
        if !fresh {
            // Composite cursor model: pointer-only motion produces NO new publish (the declared
            // hardware cursor never dirties the frame), so a static desktop would freeze the
            // blended pointer. Regenerate from the LAST slot whenever the polled cursor state
            // changed — the re-converted out-ring frame carries the pointer's new position.
            let moved = self.composite_cursor
                && self.last_slot.is_some()
                && self.cursor_blend_key() != self.last_blend_key;
            if !moved {
                return Ok(None);
            }
            slot = self.last_slot.expect("checked above");
            if slot >= self.slots.len() {
                return Ok(None); // ring shrank across a recreate — wait for a fresh publish
            }
            regen = true;
        }
        // Build the ring + converter BEFORE acquiring the slot so nothing between Acquire and Release
        // can `?`-return and leak the keyed-mutex lock (which would stall the driver on that slot).
        // PyroWave uses its OWN two-plane ring (`pyro_ring`); everything else the single NV12/BGRA ring.
@@ -1718,18 +2086,31 @@ impl IddPushCapturer {
        // Hold the slot's keyed mutex only across the convert/copy into the host out-ring (NOT across the
        // ~3 ms encode — NVENC reads the host out-ring slot, not the keyed-mutex slot), so the driver gets
        // the slot back immediately and the encode of the PREVIOUS frame overlaps this convert.
        // Clone the slot's COM interfaces (an AddRef each) so the guard borrows LOCALS, leaving
        // `self` free for the composite blend prep inside the lock.
        let (slot_tex, slot_srv, slot_mutex) = {
            let s = &self.slots[slot];
            (s.tex.clone(), s.srv.clone(), s.mutex.clone())
        };
        // Acquire the slot's keyed mutex via a RAII guard, scoped to JUST the convert/copy below so it
        // releases at the same point as the old hand-written `ReleaseSync` (the driver gets the slot back
        // immediately, NOT held across the rest of `try_consume`) — but now leak-proof on any early return.
        {
-            let Some(_lock) = KeyedMutexGuard::acquire(&s.mutex, 0, 8) else {
+            let Some(_lock) = KeyedMutexGuard::acquire(&slot_mutex, 0, 8) else {
                return Ok(None);
            };
            // SAFETY: convert on the owning (encode) thread's immediate context, holding the slot lock.
            // A `?` here is leak-safe: `_lock` (the KeyedMutexGuard) drops on the early return, releasing
            // the slot back to the driver.
            unsafe {
                // Composite cursor model: divert the convert input through the blend scratch —
                // a slot copy with the pointer quad alpha-blended on top. `None` = compositing
                // off or degraded (the conversion then reads the slot as always).
                let blended = if self.composite_cursor {
                    self.prepare_blend_scratch(&slot_tex)
                } else {
                    None
                };
                if self.pyrowave {
                    // PyroWave: ring slot SRV (BGRA for SDR, scRGB FP16 for HDR) → the two separate
                    // plane textures via the mode-aware CSC; the shared fence signalled just after
@@ -1737,22 +2118,17 @@ impl IddPushCapturer {
                    // convert. The composition format is pinned to the negotiated depth.
                    let (_, y_rtv, _, cbcr_rtv) = pyro_slot.as_ref().expect("pyro slot");
                    if let Some(conv) = self.pyro_conv.as_ref() {
-                        conv.convert(
+                        let src = blended.as_ref().map(|(_, srv)| srv).unwrap_or(&slot_srv);
-                            &self.context,
+                        conv.convert(&self.context, src, y_rtv, cbcr_rtv, self.width, self.height)?;
                            &s.srv,
                            y_rtv,
                            cbcr_rtv,
                            self.width,
                            self.height,
                        )?;
                    }
                } else if self.display_hdr {
                    // HDR: FP16 slot SRV → P010 (BT.2020 PQ) via the shader; NVENC takes native P010.
                    if let Some(conv) = self.hdr_p010_conv.as_ref() {
                        let src = blended.as_ref().map(|(_, srv)| srv).unwrap_or(&slot_srv);
                        conv.convert(
                            &self.device,
                            &self.context,
-                            &s.srv,
+                            src,
                            out.as_ref().expect("out ring"),
                            self.width,
                            self.height,
@@ -1762,12 +2138,14 @@ impl IddPushCapturer {
                    // SDR 4:4:4: pass the BGRA slot through untouched — NVENC ingests full-chroma
                    // RGB and CSCs to YUV 4:4:4 itself (per the always-written BT.709 VUI). Plain
                    // copy-engine move; the slot releases back to the driver immediately.
                    let src = blended.as_ref().map(|(t, _)| t).unwrap_or(&slot_tex);
                    self.context
-                        .CopyResource(out.as_ref().expect("out ring"), &s.tex);
+                        .CopyResource(out.as_ref().expect("out ring"), src);
                } else {
                    // SDR: BGRA slot → NV12 on the VIDEO engine; NVENC takes native NV12, no SM-side CSC.
                    if let Some(conv) = self.video_conv.as_ref() {
-                        conv.convert(&s.tex, out.as_ref().expect("out ring"))?;
+                        let src = blended.as_ref().map(|(t, _)| t).unwrap_or(&slot_tex);
                        conv.convert(src, out.as_ref().expect("out ring"))?;
                    }
                }
            }
@@ -1775,13 +2153,20 @@ impl IddPushCapturer {
        }
        self.out_idx = (i + 1) % ring_len;
        self.last_seq = seq;
        if fresh {
            self.last_slot = Some(slot);
        }
        if let Some((y, _, cbcr, _)) = pyro_slot.as_ref() {
            self.pyro_last = Some((y.clone(), cbcr.clone()));
        } else {
            self.last_present = Some((out.as_ref().expect("out ring").clone(), pf));
        }
        let now = Instant::now();
-        if self.recovering_since.take().is_some() {
+        if regen {
            // A regen re-encodes OLD desktop content at a new pointer position — it is not a
            // fresh driver frame; feeding the freshness/stall bookkeeping would mask a dead
            // driver and pollute stall attribution.
        } else if self.recovering_since.take().is_some() {
            // A fresh frame resumed → recovered. The recovery gap is self-inflicted (ring
            // recreate, already logged by the recreate path) — reset the stall watch so it
            // doesn't read as a DWM stall.
@@ -1853,7 +2238,9 @@ impl IddPushCapturer {
                }
            }
        }
        if !regen {
            self.last_fresh = now; // feeds the driver-death watch
        }
        // Build the frame. For PyroWave the encode input is the Y plane
        // (`texture`) + the CbCr plane & fence in `pyro`; signal the shared fence
        // after the convert above. SAFETY: on the owning capture/encode thread.
@@ -1981,7 +2368,82 @@ impl std::fmt::Display for AttachTexFail {
 impl std::error::Error for AttachTexFail {}
 /// Duplicate `cs`'s section into the driver's WUDFHost and send `IOCTL_SET_CURSOR_CHANNEL`.
 /// `true` = the driver adopted it (worker declared per its `cursor_forward_on` state). Shared by
 /// the open-time delivery and every RE-delivery (ring recreate / flip NOT_FOUND) — the request is
 /// idempotent driver-side (a replaced worker is stopped + joined).
 fn deliver_cursor_channel(
    broker: &ChannelBroker,
    target_id: u32,
    cs: &cursor::CursorShared,
    send_cursor: &crate::CursorChannelSender,
 ) -> bool {
    // SAFETY: `cs.section_handle()` borrows the section mapping `cs` owns (live across this
    // synchronous call); the broker's WUDFHost process handle is live for the broker's lifetime.
    let value = match unsafe { broker.dup_into_public(cs.section_handle()) } {
        Ok(v) => v,
        Err(e) => {
            tracing::warn!("cursor section duplication failed (composited cursor stays): {e:#}");
            return false;
        }
    };
    let req = pf_driver_proto::control::SetCursorChannelRequest {
        target_id,
        _pad: 0,
        header_handle: value,
    };
    match send_cursor(&req) {
        Ok(()) => {
            tracing::info!(
                target_id,
                "IDD push(host): cursor channel delivered — driver declares the hardware cursor"
            );
            true
        }
        Err(e) => {
            broker.close_remote_public(value);
            tracing::warn!("cursor channel delivery failed (composited cursor stays): {e:#}");
            false
        }
    }
 }
 impl Capturer for IddPushCapturer {
    fn cursor(&mut self) -> Option<pf_frame::CursorOverlay> {
        // A LIVE poller is the sole source — even while it still reports `None` (pre-first-shape):
        // falling back to the shm mid-session would interleave two serial namespaces and poison
        // the client's shape cache. The shm read only serves a poller that failed to start/died.
        if let Some(p) = &self.cursor_poll {
            if p.alive() {
                return p.read();
            }
        }
        self.cursor_shared.as_mut().and_then(|c| c.read())
    }
    fn set_cursor_forward(&mut self, on: bool) {
        // The composite (capture) model is implemented HOST-side: the driver's hardware cursor
        // stays declared for the session's whole life — the only dependable state (there is NO
        // working un-declare; see cursor_blend.rs) — keeping every frame pointer-free, and the
        // capturer blends the GDI poller's shape into the frame itself. No driver round-trip.
        // `composite_forced` (a channel-less session on a sticky-declared target) is pinned ON:
        // with no client drawing, un-compositing would erase the pointer entirely.
        let composite = (!on && self.cursor_shared.is_some()) || self.composite_forced;
        if self.composite_cursor != composite {
            self.composite_cursor = composite;
            self.last_blend_key = None; // regenerate immediately at the current pointer state
            tracing::info!(
                composite,
                "cursor render model: host compositing {}",
                if composite {
                    "ON (capture model — blending the pointer into frames)"
                } else {
                    "OFF (client draws locally)"
                }
            );
        }
    }
    fn next_frame(&mut self) -> Result<CapturedFrame> {
        let deadline = Instant::now() + Duration::from_secs(20);
        loop {
@@ -2115,6 +2577,16 @@ fn warn_444_hdr_downgrade_once() {
 impl Drop for IddPushCapturer {
    fn drop(&mut self) {
        // A channel session ending while the secure-desktop guard is engaged must not leave the
        // driver's per-target desired state off — the next session's channel delivery would
        // adopt UNdeclared and silently run the composite model (§8.6's cross-session trap).
        // The open-time reset also covers this (host-crash case); this is the orderly-teardown
        // belt.
        if self.secure_active && self.cursor_shared.is_some() {
            if let Some(fwd) = self.cursor_forward.as_ref() {
                let _ = fwd(true);
            }
        }
        self.slots.clear();
        // The shared header section (`MappedSection`), the frame-ready `event` (`OwnedHandle`) and the
        // broker's WUDFHost process handle free themselves via RAII (unmap view, then close handle) —
@@ -105,6 +105,22 @@ impl ChannelBroker {
        Ok(out.0 as usize as u64)
    }
    /// Duplicate the cursor section into WUDFHost (v5 cursor channel) with the same
    /// least-privilege section rights as the frame header. Thin `pub(super)` face over
    /// [`dup_into`](Self::dup_into) for the cursor-delivery path in `open_on`.
    ///
    /// # Safety
    /// `h` must be a live handle of the current process.
    pub(super) unsafe fn dup_into_public(&self, h: HANDLE) -> Result<u64> {
        // SAFETY: forwarded contract — `h` is live per this fn's own contract.
        unsafe { self.dup_into(h, Some(SECTION_MAP_RW)) }
    }
    /// [`close_remote`](Self::close_remote) for the cursor-delivery failure path.
    pub(super) fn close_remote_public(&self, value: u64) {
        self.close_remote(value);
    }
    /// Close a handle VALUE inside the WUDFHost table (the failure-path reaper): `DUPLICATE_CLOSE_SOURCE`
    /// with no target closes the source handle regardless of the (ignored) result.
    fn close_remote(&self, value: u64) {
@@ -0,0 +1,194 @@
 //! Host side of the v5 hardware-cursor channel (remote-desktop-sweep M2c): the capturer creates
 //! an unnamed [`CursorShm`] section, delivers it to the pf-vdisplay driver (which declares an
 //! IddCx hardware cursor — DWM then EXCLUDES the pointer from the frames we consume), and reads
 //! the driver's seqlock publishes here at encode-tick pace, converting them into the same
 //! [`pf_frame::CursorOverlay`] the Linux portal path produces — everything downstream (the
 //! cursor forwarder, the wire, the client renderer) is shared.
 // Every `unsafe` block in this file carries a `// SAFETY:` proof; enforce it.
 #![deny(clippy::undocumented_unsafe_blocks)]
 use super::*;
 use pf_driver_proto::cursor::{
    CursorShm, CURSOR_MAGIC, CURSOR_SHAPE_BYTES, CURSOR_SHAPE_MAX, CURSOR_SHAPE_OFFSET,
    CURSOR_SHM_SIZE, CURSOR_TYPE_MASKED_COLOR,
 };
 use std::sync::atomic::AtomicU32;
 /// The host end of one monitor's cursor channel: the section (we created it — the mapping stays
 /// valid for the capturer's life) plus the reader's conversion cache.
 pub(super) struct CursorShared {
    section: MappedSection,
    /// The monitor's desktop origin — IddCx reports positions in DESKTOP coordinates; the
    /// overlay wants frame-relative. Fetched at attach (the virtual monitor's placement is
    /// stable for the session; a topology change recreates the pipeline anyway).
    origin: (i32, i32),
    /// Conversion cache: the last `shape_id` whose pixels were converted, and the result.
    /// Position-only updates (the common case) reuse it — a refcount bump, no pixel work.
    cached_id: u32,
    cached: Option<ConvertedShape>,
 }
 struct ConvertedShape {
    rgba: std::sync::Arc<Vec<u8>>,
    w: u32,
    h: u32,
    hot_x: u32,
    hot_y: u32,
 }
 impl CursorShared {
    /// Create + initialize the section (magic stamped, seq even/zero). The returned handle is
    /// the section itself (owned by `self`); the caller duplicates it into the WUDFHost.
    pub(super) fn create(target_id: u32) -> Result<CursorShared> {
        // SAFETY: plain FFI. Unnamed pagefile-backed section, host-lifetime owned; the view is
        // mapped once and unmapped never (the capturer's life = the session's life).
        let section = unsafe {
            let map = CreateFileMappingW(
                INVALID_HANDLE_VALUE,
                None,
                PAGE_READWRITE,
                0,
                CURSOR_SHM_SIZE as u32,
                PCWSTR::null(),
            )
            .context("CreateFileMapping(cursor)")?;
            let map = OwnedHandle::from_raw_handle(map.0 as _);
            let view = MapViewOfFile(
                HANDLE(map.as_raw_handle()),
                FILE_MAP_ALL_ACCESS,
                0,
                0,
                CURSOR_SHM_SIZE,
            );
            if view.Value.is_null() {
                bail!("MapViewOfFile failed for the cursor section");
            }
            let shm = view.Value.cast::<CursorShm>();
            std::ptr::write_bytes(view.Value.cast::<u8>(), 0, CURSOR_SHM_SIZE);
            // Magic LAST-ish (the driver validates it at adopt; seq 0 = even = consistent).
            std::sync::atomic::fence(Ordering::Release);
            (*shm).magic = CURSOR_MAGIC;
            MappedSection { handle: map, view }
        };
        // Desktop origin of this monitor's source — for the desktop→frame coordinate shift.
        // SAFETY: `source_desktop_rect` only runs the CCD QueryDisplayConfig FFI over owned
        // locals (same call the compose-kick path makes).
        let rect = unsafe { pf_win_display::win_display::source_desktop_rect(target_id) };
        let origin = rect.map(|(x, y, _w, _h)| (x, y)).unwrap_or((0, 0));
        Ok(CursorShared {
            section,
            origin,
            cached_id: 0,
            cached: None,
        })
    }
    /// The section handle for the broker's duplication into the WUDFHost.
    pub(super) fn section_handle(&self) -> HANDLE {
        HANDLE(self.section.handle.as_raw_handle())
    }
    /// Seqlock-read the driver's latest publish → a frame-relative [`pf_frame::CursorOverlay`].
    /// `None` until the first publish lands (or while the pointer has never been seen). A hidden
    /// pointer returns `Some` with `visible: false` — the forwarder turns that into the client's
    /// relative-mode hint, exactly like the Linux path.
    pub(super) fn read(&mut self) -> Option<pf_frame::CursorOverlay> {
        let shm = self.section.ptr::<CursorShm>();
        // SAFETY: the view spans CURSOR_SHM_SIZE for self's lifetime; seq is 4-aligned in the
        // fixed layout (offset 4).
        let seq = unsafe { &*std::ptr::addr_of!((*shm).seq).cast::<AtomicU32>() };
        for _ in 0..64 {
            let s1 = seq.load(Ordering::Acquire);
            if s1 == 0 {
                return None; // no publish yet
            }
            if s1 & 1 != 0 {
                std::hint::spin_loop();
                continue; // writer mid-update
            }
            // SAFETY: header reads within the mapped view; consistency is validated by the
            // seq re-check below (a torn read is discarded and retried).
            let hdr = unsafe { std::ptr::read_volatile(shm) };
            // Shape pixels: convert only when the OS minted a new shape id.
            if hdr.visible != 0 && hdr.shape_id != self.cached_id {
                let rows = hdr.height.min(CURSOR_SHAPE_MAX) as usize;
                let width = hdr.width.min(CURSOR_SHAPE_MAX) as usize;
                let pitch = (hdr.pitch as usize).min(CURSOR_SHAPE_BYTES / rows.max(1));
                let mut raw = vec![0u8; rows * pitch];
                // SAFETY: the shape region spans CURSOR_SHAPE_BYTES from CURSOR_SHAPE_OFFSET
                // inside the mapped view; `rows * pitch` is clamped to it above.
                unsafe {
                    std::ptr::copy_nonoverlapping(
                        self.section.ptr::<u8>().add(CURSOR_SHAPE_OFFSET),
                        raw.as_mut_ptr(),
                        rows * pitch,
                    );
                }
                // Discard the copy if the writer raced us mid-shape (seq moved) — retry.
                if seq.load(Ordering::Acquire) != s1 {
                    continue;
                }
                self.cached = Some(convert_shape(&hdr, &raw, width, rows, pitch));
                self.cached_id = hdr.shape_id;
            } else if seq.load(Ordering::Acquire) != s1 {
                continue;
            }
            let shape = self.cached.as_ref()?;
            return Some(pf_frame::CursorOverlay {
                x: hdr.x - self.origin.0,
                y: hdr.y - self.origin.1,
                w: shape.w,
                h: shape.h,
                rgba: shape.rgba.clone(),
                serial: u64::from(hdr.shape_id),
                hot_x: shape.hot_x,
                hot_y: shape.hot_y,
                visible: hdr.visible != 0,
            });
        }
        None // persistent tearing (writer wedged mid-seq) — skip this tick
    }
 }
 /// Convert the OS's 32-bpp pitch-strided shape rows into the overlay's packed straight RGBA.
 /// ALPHA cursors are BGRA with straight per-pixel alpha (swap R↔B). MASKED_COLOR approximates:
 /// alpha 0x00 = opaque color pixel; 0xFF = an XOR pixel we cannot honor client-side — rendered
 /// as a translucent mid-gray so inversion cursors stay visible instead of vanishing.
 fn convert_shape(
    hdr: &CursorShm,
    raw: &[u8],
    width: usize,
    rows: usize,
    pitch: usize,
 ) -> ConvertedShape {
    let masked = hdr.cursor_type == CURSOR_TYPE_MASKED_COLOR;
    let mut rgba = Vec::with_capacity(width * rows * 4);
    for y in 0..rows {
        let row = &raw[y * pitch..];
        for x in 0..width {
            let o = x * 4;
            if o + 4 > row.len() {
                rgba.extend_from_slice(&[0, 0, 0, 0]);
                continue;
            }
            let (b, g, r, a) = (row[o], row[o + 1], row[o + 2], row[o + 3]);
            if masked {
                if a == 0 {
                    rgba.extend_from_slice(&[r, g, b, 0xFF]);
                } else {
                    rgba.extend_from_slice(&[0x80, 0x80, 0x80, 0xB4]);
                }
            } else {
                rgba.extend_from_slice(&[r, g, b, a]);
            }
        }
    }
    ConvertedShape {
        rgba: std::sync::Arc::new(rgba),
        w: width as u32,
        h: rows as u32,
        hot_x: hdr.hot_x.min(width.saturating_sub(1) as u32),
        hot_y: hdr.hot_y.min(rows.saturating_sub(1) as u32),
    }
 }
@@ -0,0 +1,216 @@
 //! Host-side cursor compositing for the CAPTURE mouse model (design/remote-desktop-sweep.md §8).
 //!
 //! Why the host draws it: once a monitor has ever declared an IddCx hardware cursor, DWM will
 //! not composite the software cursor back into its frames — there is no un-declare DDI (the
 //! empty-caps re-setup is rejected `STATUS_INVALID_PARAMETER`), and a successful same-mode
 //! re-commit with the driver's re-declare provably suppressed still leaves the pointer excluded
 //! (all observed on-glass, 26100). So the driver keeps its hardware cursor declared for the
 //! session's whole life — the state that works — and when the client flips to the capture model
 //! the HOST composites the pointer into the frame itself: a slot→scratch copy plus one
 //! alpha-blended quad (the GDI poller's full-fidelity shape at its polled position), entirely
 //! GPU-side on the capture device, before the normal conversion runs from the scratch.
 // Every `unsafe` block in this file carries a `// SAFETY:` proof; enforce it (unsafe-proof program).
 #![deny(clippy::undocumented_unsafe_blocks)]
 use super::*;
 use windows::core::s;
 use windows::Win32::Graphics::Direct3D::D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST;
 use windows::Win32::Graphics::Direct3D11::{
    ID3D11BlendState, ID3D11Buffer, ID3D11PixelShader, ID3D11SamplerState, ID3D11VertexShader,
    D3D11_BIND_CONSTANT_BUFFER, D3D11_BLEND_DESC, D3D11_BLEND_INV_SRC_ALPHA, D3D11_BLEND_ONE,
    D3D11_BLEND_OP_ADD, D3D11_BLEND_SRC_ALPHA, D3D11_BUFFER_DESC, D3D11_COMPARISON_NEVER,
    D3D11_CPU_ACCESS_WRITE, D3D11_FILTER_MIN_MAG_MIP_LINEAR, D3D11_MAPPED_SUBRESOURCE,
    D3D11_MAP_WRITE_DISCARD, D3D11_RENDER_TARGET_BLEND_DESC, D3D11_SAMPLER_DESC,
    D3D11_SUBRESOURCE_DATA, D3D11_TEXTURE_ADDRESS_CLAMP, D3D11_USAGE_DYNAMIC, D3D11_VIEWPORT,
 };
 use windows::Win32::Graphics::Dxgi::Common::DXGI_FORMAT_R8G8B8A8_UNORM;
 /// Straight-alpha sample of the cursor bitmap. `linear_scale` = 0 passes sRGB through (SDR
 /// ring); non-zero linearizes sRGB→scRGB AND multiplies by the target's SDR-white scale
 /// (`sdr_white_level_scale` — 1.0 would put cursor-white at 80 nits, visibly DARKER than the
 /// surrounding SDR desktop content DWM composes at the user's SDR-brightness setting).
 const CURSOR_PS: &str = r"
 Texture2D<float4> tx : register(t0);
 SamplerState sm : register(s0);
 cbuffer C : register(b0) { float linear_scale; float3 pad; };
 float4 main(float4 pos : SV_POSITION, float2 uv : TEXCOORD0) : SV_Target {
    float4 c = tx.Sample(sm, uv);
    if (linear_scale != 0.0) {
        c.rgb = pow(abs(c.rgb), 2.2) * linear_scale;
    }
    return c;
 }
 ";
 /// The cursor-quad blend pass + its shape-texture cache. One per capturer (device-scoped).
 pub(super) struct CursorBlendPass {
    vs: ID3D11VertexShader,
    ps: ID3D11PixelShader,
    sampler: ID3D11SamplerState,
    blend: ID3D11BlendState,
    cbuf: ID3D11Buffer,
    cbuf_scale: Option<f32>,
    /// The uploaded shape (serial-keyed): SRV + dims in host pixels.
    shape: Option<(u64, ID3D11ShaderResourceView, u32, u32)>,
 }
 impl CursorBlendPass {
    pub(super) unsafe fn new(device: &ID3D11Device) -> Result<Self> {
        let vsb = crate::dxgi::compile_shader(crate::dxgi::HDR_VS, s!("main"), s!("vs_5_0"))?;
        let psb = crate::dxgi::compile_shader(CURSOR_PS, s!("main"), s!("ps_5_0"))?;
        let mut vs = None;
        device.CreateVertexShader(&vsb, None, Some(&mut vs))?;
        let mut ps = None;
        device.CreatePixelShader(&psb, None, Some(&mut ps))?;
        let sd = D3D11_SAMPLER_DESC {
            // LINEAR: the quad is drawn 1:1 in frame pixels, so this only matters at the
            // half-texel edges; linear keeps them soft instead of ringing.
            Filter: D3D11_FILTER_MIN_MAG_MIP_LINEAR,
            AddressU: D3D11_TEXTURE_ADDRESS_CLAMP,
            AddressV: D3D11_TEXTURE_ADDRESS_CLAMP,
            AddressW: D3D11_TEXTURE_ADDRESS_CLAMP,
            ComparisonFunc: D3D11_COMPARISON_NEVER,
            MaxLOD: f32::MAX,
            ..Default::default()
        };
        let mut sampler = None;
        device.CreateSamplerState(&sd, Some(&mut sampler))?;
        // Straight-alpha over: dst.rgb = src.rgb*a + dst.rgb*(1-a); keep dst alpha.
        let mut bd = D3D11_BLEND_DESC::default();
        bd.RenderTarget[0] = D3D11_RENDER_TARGET_BLEND_DESC {
            BlendEnable: true.into(),
            SrcBlend: D3D11_BLEND_SRC_ALPHA,
            DestBlend: D3D11_BLEND_INV_SRC_ALPHA,
            BlendOp: D3D11_BLEND_OP_ADD,
            SrcBlendAlpha: D3D11_BLEND_ONE,
            DestBlendAlpha: D3D11_BLEND_ONE,
            BlendOpAlpha: D3D11_BLEND_OP_ADD,
            RenderTargetWriteMask: 0x0F,
        };
        let mut blend = None;
        device.CreateBlendState(&bd, Some(&mut blend))?;
        let cbd = D3D11_BUFFER_DESC {
            ByteWidth: 16, // float to_linear + float3 pad
            Usage: D3D11_USAGE_DYNAMIC,
            BindFlags: D3D11_BIND_CONSTANT_BUFFER.0 as u32,
            CPUAccessFlags: D3D11_CPU_ACCESS_WRITE.0 as u32,
            ..Default::default()
        };
        let mut cbuf = None;
        device.CreateBuffer(&cbd, None, Some(&mut cbuf))?;
        Ok(Self {
            vs: vs.context("cursor blend vs")?,
            ps: ps.context("cursor blend ps")?,
            sampler: sampler.context("cursor blend sampler")?,
            blend: blend.context("cursor blend state")?,
            cbuf: cbuf.context("cursor blend cbuf")?,
            cbuf_scale: None,
            shape: None,
        })
    }
    /// Upload `ov`'s bitmap if its serial is new; reuse the cached SRV otherwise.
    unsafe fn ensure_shape(
        &mut self,
        device: &ID3D11Device,
        ov: &pf_frame::CursorOverlay,
    ) -> Result<()> {
        if self.shape.as_ref().is_some_and(|(s, ..)| *s == ov.serial) {
            return Ok(());
        }
        if ov.rgba.len() < (ov.w as usize) * (ov.h as usize) * 4 || ov.w == 0 || ov.h == 0 {
            bail!("malformed cursor overlay ({}x{})", ov.w, ov.h);
        }
        let desc = D3D11_TEXTURE2D_DESC {
            Width: ov.w,
            Height: ov.h,
            MipLevels: 1,
            ArraySize: 1,
            Format: DXGI_FORMAT_R8G8B8A8_UNORM,
            SampleDesc: DXGI_SAMPLE_DESC {
                Count: 1,
                Quality: 0,
            },
            Usage: D3D11_USAGE_DEFAULT,
            BindFlags: D3D11_BIND_SHADER_RESOURCE.0 as u32,
            ..Default::default()
        };
        let init = D3D11_SUBRESOURCE_DATA {
            pSysMem: ov.rgba.as_ptr().cast(),
            SysMemPitch: ov.w * 4,
            SysMemSlicePitch: 0,
        };
        let mut tex: Option<ID3D11Texture2D> = None;
        device
            .CreateTexture2D(&desc, Some(&init), Some(&mut tex))
            .context("CreateTexture2D(cursor shape)")?;
        let tex = tex.context("null cursor shape texture")?;
        let mut srv: Option<ID3D11ShaderResourceView> = None;
        device
            .CreateShaderResourceView(&tex, None, Some(&mut srv))
            .context("CreateShaderResourceView(cursor shape)")?;
        self.shape = Some((ov.serial, srv.context("null cursor shape srv")?, ov.w, ov.h));
        Ok(())
    }
    /// Alpha-blend `ov` onto `dst` (frame-sized, RENDER_TARGET-capable — the blend scratch).
    /// `linear_scale`: 0 = SDR passthrough; non-zero = the frame is FP16 scRGB (HDR
    /// composition) — linearize and scale to the target's SDR white. The quad is placed purely
    /// via the viewport (the fullscreen-triangle VS fills whatever viewport is set), clipped by
    /// the target automatically.
    pub(super) unsafe fn blend(
        &mut self,
        device: &ID3D11Device,
        ctx: &ID3D11DeviceContext,
        dst: &ID3D11Texture2D,
        ov: &pf_frame::CursorOverlay,
        linear_scale: f32,
    ) -> Result<()> {
        self.ensure_shape(device, ov)?;
        let (_, srv, w, h) = self.shape.as_ref().expect("shape just ensured");
        if self.cbuf_scale != Some(linear_scale) {
            let cb: [f32; 4] = [linear_scale, 0.0, 0.0, 0.0];
            let mut mapped = D3D11_MAPPED_SUBRESOURCE::default();
            if ctx
                .Map(&self.cbuf, 0, D3D11_MAP_WRITE_DISCARD, 0, Some(&mut mapped))
                .is_ok()
            {
                std::ptr::copy_nonoverlapping(cb.as_ptr(), mapped.pData as *mut f32, cb.len());
                ctx.Unmap(&self.cbuf, 0);
            }
            self.cbuf_scale = Some(linear_scale);
        }
        let mut rtv: Option<ID3D11RenderTargetView> = None;
        device
            .CreateRenderTargetView(dst, None, Some(&mut rtv))
            .context("CreateRenderTargetView(cursor blend scratch)")?;
        let rtv = rtv.context("null cursor blend rtv")?;
        ctx.OMSetRenderTargets(Some(&[Some(rtv)]), None);
        ctx.OMSetBlendState(&self.blend, None, 0xffff_ffff);
        ctx.VSSetShader(&self.vs, None);
        ctx.PSSetShader(&self.ps, None);
        ctx.PSSetShaderResources(0, Some(&[Some(srv.clone())]));
        ctx.PSSetSamplers(0, Some(&[Some(self.sampler.clone())]));
        ctx.PSSetConstantBuffers(0, Some(&[Some(self.cbuf.clone())]));
        ctx.IASetInputLayout(None);
        ctx.IASetPrimitiveTopology(D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST);
        // Placement IS the viewport: the VS fills it, the OS clips it to the target.
        let vp = D3D11_VIEWPORT {
            TopLeftX: ov.x as f32,
            TopLeftY: ov.y as f32,
            Width: *w as f32,
            Height: *h as f32,
            MinDepth: 0.0,
            MaxDepth: 1.0,
        };
        ctx.RSSetViewports(Some(&[vp]));
        ctx.Draw(3, 0);
        // Unbind so the scratch can be bound as a conversion INPUT without a hazard warning.
        ctx.OMSetRenderTargets(None, None);
        let none_srv: [Option<ID3D11ShaderResourceView>; 1] = [None];
        ctx.PSSetShaderResources(0, Some(&none_srv));
        Ok(())
    }
 }
@@ -0,0 +1,508 @@
 //! GDI cursor poller — the Windows cursor-SHAPE source for the cursor-forward channel
 //! (design/remote-desktop-sweep.md §8, the M2c redesign).
 //!
 //! Why not the IddCx hardware-cursor query (the v5 `CursorShm` path, now the fallback): it is
 //! alpha-only BY DESIGN — `IDDCX_CURSOR_SHAPE_TYPE` has no monochrome value, the OS pre-converts
 //! monochrome to masked-color (IDDCX_CURSOR_CAPS docs), and masked-color delivery is dead code on
 //! modern builds (proven on-glass at every `ColorXorCursorSupport` level; no public evidence of a
 //! MASKED_COLOR delivery anywhere). The driver keeps its hardware cursor declared at XOR FULL
 //! purely so DWM EXCLUDES every cursor type from the IDD frame.
 //!
 //! Why not DXGI Desktop Duplication `GetFramePointerShape`: its `PointerPosition.Visible` goes
 //! stale when the cursor moves only via injected input on current Win11 (Sunshine #5293 — exactly
 //! a Punktfunk session's topology), it burns one of the session's four duplication slots, and its
 //! per-output metadata on IDD monitors has conflicting field reports. The GDI path below is the
 //! metadata-forwarding-remote-desktop pattern (RustDesk, WebRTC/Chrome Remote Desktop, OBS): the
 //! cursor is per-session global state in win32k, readable cross-process, and `CURSOR_SHOWING` is
 //! the logical visibility — immune to all of the above.
 //!
 //! Works because the capture host runs as SYSTEM *inside the interactive session* on
 //! `winsta0\default` (the service supervisor retargets the token — `windows/service.rs`
 //! `spawn_host`), so the poller thread sees the session's cursor directly; no helper process.
 // Every `unsafe` block in this file carries a `// SAFETY:` proof; enforce it (unsafe-proof program).
 #![deny(clippy::undocumented_unsafe_blocks)]
 use super::*;
 use windows::Win32::Graphics::Gdi::{
    DeleteObject, GetDC, GetDIBits, GetObjectW, ReleaseDC, BITMAP, BITMAPINFO, BITMAPINFOHEADER,
    BI_RGB, DIB_RGB_COLORS, HBITMAP, HDC,
 };
 use windows::Win32::System::StationsAndDesktops::{
    CloseDesktop, GetUserObjectInformationW, OpenInputDesktop, SetThreadDesktop,
    DESKTOP_ACCESS_FLAGS, DESKTOP_CONTROL_FLAGS, HDESK, UOI_NAME,
 };
 use windows::Win32::UI::HiDpi::{
    SetThreadDpiAwarenessContext, DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2,
 };
 use windows::Win32::UI::WindowsAndMessaging::{
    CopyIcon, DestroyIcon, GetCursorInfo, GetIconInfo, CURSORINFO, HICON, ICONINFO,
 };
 /// `CURSORINFO.flags` bits (WindowsAndMessaging): the pointer is logically shown /
 /// touch-or-pen-suppressed. Named locally so the visibility rule below reads as the docs do.
 const CURSOR_SHOWING: u32 = 0x1;
 const CURSOR_SUPPRESSED: u32 = 0x2;
 /// A converted shape: the cache the per-tick overlay is assembled from. `rgba` is `Arc` so the
 /// slot publish (and every downstream frame attach) is a refcount bump.
 struct Shape {
    rgba: std::sync::Arc<Vec<u8>>,
    w: u32,
    h: u32,
    hot_x: u32,
    hot_y: u32,
    serial: u64,
 }
 /// Off-thread GDI cursor poller. Samples `GetCursorInfo` at ~60 Hz, rasterises the `HCURSOR` only
 /// when its handle value changes, and publishes a ready [`pf_frame::CursorOverlay`] snapshot; the
 /// capture thread's per-tick cost is one uncontended mutex read + an `Arc` clone
 /// (same split as [`DescriptorPoller`], and for the same reason: user32/gdi32 calls have no place
 /// on the capture/encode thread).
 pub(super) struct CursorPoller {
    slot: Arc<Mutex<Option<pf_frame::CursorOverlay>>>,
    stop: Arc<AtomicBool>,
    /// The input desktop is a SECURE desktop (Winlogon — UAC consent / lock / logon). Classified
    /// on every reattach; the capturer polls it to stand the IddCx hardware-cursor declare down
    /// while the secure desktop needs the software-cursor path to render (see
    /// `IddPushCapturer::poll_secure_desktop`).
    secure: Arc<AtomicBool>,
    thread: Option<std::thread::JoinHandle<()>>,
 }
 impl CursorPoller {
    /// ~250 Hz: the polled position is ALSO the composite-blend position (capture model), so
    /// it must out-pace the fastest session — at 16 ms a 240 fps stream re-used a stale
    /// position for ~4 consecutive frames and the composited pointer visibly stuttered
    /// against the video. A tick is one `GetCursorInfo` syscall (rasterisation only on shape
    /// change), so 250 Hz is still negligible CPU.
    const INTERVAL: Duration = Duration::from_millis(4);
    /// Unconditional input-desktop reattach cadence — catches secure-desktop (UAC/lock) switches
    /// without a failure signal (`GetCursorInfo` on a stale desktop *succeeds* with stale data).
    /// 250 ms, not the original 2 s: the reattach now also feeds [`Self::secure_desktop`], which
    /// gates when the secure desktop becomes VISIBLE in the stream (the hardware-cursor
    /// stand-down) — a 2 s freeze at every UAC prompt is user-visible, ~4 `OpenInputDesktop`
    /// syscalls/s are not.
    const REATTACH: Duration = Duration::from_millis(250);
    /// Spawn the poller for the virtual display `target_id`. `rect` = the target's desktop rect
    /// (`source_desktop_rect` order: x, y, w, h) — cursor positions are desktop-global; the
    /// overlay wants frame-relative, and a pointer outside the rect reports `visible: false`
    /// (per-output semantics, matching the driver shm path and the Linux portal).
    pub(super) fn spawn(target_id: u32, rect: (i32, i32, i32, i32)) -> Self {
        let slot: Arc<Mutex<Option<pf_frame::CursorOverlay>>> = Arc::new(Mutex::new(None));
        let stop = Arc::new(AtomicBool::new(false));
        let secure = Arc::new(AtomicBool::new(false));
        let (slot_t, stop_t, secure_t) = (slot.clone(), stop.clone(), secure.clone());
        let thread = std::thread::Builder::new()
            .name("pf-cursor-poll".into())
            .spawn(move || run(target_id, rect, &slot_t, &stop_t, &secure_t))
            .ok();
        if thread.is_none() {
            tracing::warn!("cursor poller thread spawn failed — cursor falls back to driver shm");
        }
        Self {
            slot,
            stop,
            secure,
            thread,
        }
    }
    /// The latest overlay snapshot (`None` until the first successful shape rasterisation).
    pub(super) fn read(&self) -> Option<pf_frame::CursorOverlay> {
        self.slot.lock().unwrap_or_else(|p| p.into_inner()).clone()
    }
    /// Whether the input desktop is currently a SECURE desktop (UAC consent / Winlogon lock or
    /// logon). Latched by the poll thread on its reattach cadence (≤ [`Self::REATTACH`] stale).
    pub(super) fn secure_desktop(&self) -> bool {
        self.secure.load(Ordering::Relaxed)
    }
    /// Whether the worker thread is (still) alive — `false` degrades the capturer to the shm read.
    pub(super) fn alive(&self) -> bool {
        self.thread.as_ref().is_some_and(|t| !t.is_finished())
    }
 }
 impl Drop for CursorPoller {
    fn drop(&mut self) {
        self.stop.store(true, Ordering::Relaxed);
        if let Some(t) = self.thread.take() {
            let _ = t.join(); // worker sleeps ≤ INTERVAL — a bounded join
        }
    }
 }
 /// The poll loop. Owns the thread's input-desktop binding and the shape cache.
 fn run(
    target_id: u32,
    rect: (i32, i32, i32, i32),
    slot: &Mutex<Option<pf_frame::CursorOverlay>>,
    stop: &AtomicBool,
    secure: &AtomicBool,
 ) {
    // Physical-pixel coordinates on this thread regardless of the process's DPI awareness:
    // `rect` comes from CCD (always physical), and a DPI-virtualized `GetCursorInfo` position
    // would land in the wrong frame pixel on any scaled display. Thread-scoped, so the rest of
    // the host is untouched.
    // SAFETY: takes and returns only a by-value context handle; affects this thread only.
    let _ = unsafe { SetThreadDpiAwarenessContext(DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2) };
    let mut desktop = DesktopBinding::default();
    // best-effort: already on winsta0\default if this fails
    publish_secure(secure, desktop.reattach());
    let mut last_attach = Instant::now();
    let mut shape: Option<Shape> = None;
    let mut cached_handle: isize = 0;
    let mut failed_handle: isize = 0; // don't re-rasterise a failing handle every tick
    let mut serial: u64 = 0;
    let mut logged_live = false;
    while !stop.load(Ordering::Relaxed) {
        std::thread::sleep(CursorPoller::INTERVAL);
        if last_attach.elapsed() >= CursorPoller::REATTACH {
            last_attach = Instant::now();
            publish_secure(secure, desktop.reattach());
        }
        let mut ci = CURSORINFO {
            cbSize: std::mem::size_of::<CURSORINFO>() as u32,
            ..Default::default()
        };
        // SAFETY: `ci` is a live, correctly-sized out-param for this synchronous call; no pointer
        // escapes it.
        if unsafe { GetCursorInfo(&mut ci) }.is_err() {
            // Desktop went away under us (secure-desktop switch mid-call) — rebind and retry
            // next tick; the slot keeps its last snapshot meanwhile.
            publish_secure(secure, desktop.reattach());
            last_attach = Instant::now();
            continue;
        }
        let flags = ci.flags.0;
        let showing = flags & CURSOR_SHOWING != 0 && flags & CURSOR_SUPPRESSED == 0;
        // Rasterise on handle change only (position-only ticks are a header update). Hidden
        // cursors keep the cached shape — the forwarder's hidden-but-known contract needs the
        // bitmap to have been seen. v1: animated cursors publish their first frame (the OBS
        // behavior); frame cycling via DrawIconEx istep is a known follow-up.
        let handle = ci.hCursor.0 as isize;
        if showing && handle != 0 && handle != cached_handle && handle != failed_handle {
            match rasterize(ci.hCursor) {
                Some((rgba, w, h, hot_x, hot_y)) => {
                    serial += 1;
                    shape = Some(Shape {
                        rgba: std::sync::Arc::new(rgba),
                        w,
                        h,
                        hot_x,
                        hot_y,
                        serial,
                    });
                    cached_handle = handle;
                    failed_handle = 0;
                    if !logged_live {
                        logged_live = true;
                        tracing::info!(
                            target_id,
                            "cursor poller live — GDI shape source publishing (serial 1: {w}x{h})"
                        );
                    }
                }
                None => {
                    // The owning app may have destroyed the cursor mid-read; keep the previous
                    // shape and don't hammer this handle again until it changes.
                    failed_handle = handle;
                }
            }
        }
        let overlay = shape.as_ref().map(|s| {
            let (px, py) = (ci.ptScreenPos.x - rect.0, ci.ptScreenPos.y - rect.1);
            let in_rect = px >= 0 && py >= 0 && px < rect.2 && py < rect.3;
            pf_frame::CursorOverlay {
                // Overlay x/y = bitmap top-left (reported position − hotspot), frame pixels.
                x: px - s.hot_x as i32,
                y: py - s.hot_y as i32,
                w: s.w,
                h: s.h,
                rgba: s.rgba.clone(),
                serial: s.serial,
                hot_x: s.hot_x,
                hot_y: s.hot_y,
                visible: showing && in_rect,
            }
        });
        *slot.lock().unwrap_or_else(|p| p.into_inner()) = overlay;
    }
 }
 /// Store a reattach's secure-desktop verdict (`None` = classification unavailable — keep the
 /// previous state rather than flapping the capturer's hardware-cursor stand-down).
 fn publish_secure(secure: &AtomicBool, verdict: Option<bool>) {
    if let Some(s) = verdict {
        secure.store(s, Ordering::Relaxed);
    }
 }
 /// The thread's owned input-desktop handle — the [`SendInputInjector`] reattach model
 /// (`pf-inject` sendinput.rs): keep the current binding, swap on demand, close exactly once.
 #[derive(Default)]
 struct DesktopBinding(Option<HDESK>);
 impl DesktopBinding {
    /// Rebind to the CURRENT input desktop. Returns whether that desktop is a SECURE one
    /// (`UOI_NAME` != "Default": "Winlogon" during UAC consent / lock / logon) — `None` when the
    /// input desktop could not be opened, in which case the binding (and the caller's secure
    /// state) stays put.
    fn reattach(&mut self) -> Option<bool> {
        const GENERIC_ALL: u32 = 0x1000_0000;
        // SAFETY: `OpenInputDesktop`/`SetThreadDesktop`/`CloseDesktop` take only by-value args.
        // `OpenInputDesktop` yields an owned `HDESK` only on `Ok`; it is either installed (and the
        // previously-owned handle closed exactly once) or closed on failure — no handle is leaked
        // or used after close. `SetThreadDesktop` rebinds only this calling thread (which owns
        // no windows/hooks, so the rebind cannot fail on that account).
        unsafe {
            match OpenInputDesktop(
                DESKTOP_CONTROL_FLAGS(0),
                false,
                DESKTOP_ACCESS_FLAGS(GENERIC_ALL),
            ) {
                Ok(h) => {
                    let secure = desktop_is_secure(h);
                    if SetThreadDesktop(h).is_ok() {
                        if let Some(old) = self.0.replace(h) {
                            let _ = CloseDesktop(old);
                        }
                    } else {
                        let _ = CloseDesktop(h);
                    }
                    Some(secure)
                }
                Err(_) => None, // not privileged for this desktop; stay put
            }
        }
    }
 }
 /// `UOI_NAME` of `h` != "Default" — i.e. the input desktop is Winlogon (UAC consent / lock /
 /// logon) or a screen-saver desktop, both of which need the OS's software-cursor render path.
 /// Unnameable desktops read as NOT secure: the only in-contract failure is a too-small buffer,
 /// and misreading secure-as-normal merely keeps today's behavior for a beat.
 fn desktop_is_secure(h: HDESK) -> bool {
    let mut name = [0u16; 64]; // "Default"/"Winlogon"/"Screen-saver" all fit with room to spare
    let mut needed = 0u32;
    // SAFETY: `h` is the live desktop handle the caller just opened; `name`/`needed` are live
    // out-params sized exactly as passed; the call writes at most `nlength` bytes.
    let ok = unsafe {
        GetUserObjectInformationW(
            windows::Win32::Foundation::HANDLE(h.0),
            UOI_NAME,
            Some(name.as_mut_ptr().cast()),
            (name.len() * 2) as u32,
            Some(&mut needed),
        )
    };
    if ok.is_err() {
        return false;
    }
    let len = name.iter().position(|&c| c == 0).unwrap_or(name.len());
    let name = String::from_utf16_lossy(&name[..len]);
    !name.eq_ignore_ascii_case("Default")
 }
 impl Drop for DesktopBinding {
    fn drop(&mut self) {
        if let Some(h) = self.0.take() {
            // SAFETY: `h` is our owned desktop handle, closed exactly once here.
            let _ = unsafe { CloseDesktop(h) };
        }
    }
 }
 /// Rasterise `hcursor` to straight-alpha RGBA: `(rgba, w, h, hot_x, hot_y)`. `None` on any
 /// failure (caller keeps the previous shape).
 fn rasterize(hcursor: windows::Win32::UI::WindowsAndMessaging::HCURSOR) -> RasterOut {
    // CopyIcon first: the owning process can destroy its HCURSOR between GetCursorInfo and the
    // reads below; the copy is ours (the OBS/WebRTC guard).
    // SAFETY: `HICON(hcursor.0)` reinterprets the cursor handle as an icon handle (cursors ARE
    // icons in user32); CopyIcon yields an owned HICON we destroy below.
    let Ok(icon) = (unsafe { CopyIcon(HICON(hcursor.0)) }) else {
        return None;
    };
    let mut ii = ICONINFO::default();
    // SAFETY: `ii` is a live out-param. On Ok it hands us COPIES of the mask/color bitmaps —
    // both deleted below (GDI-handle leak otherwise).
    let got = unsafe { GetIconInfo(icon, &mut ii) };
    let out = if got.is_ok() { convert(&ii) } else { None };
    // SAFETY: deleting the two bitmap copies GetIconInfo returned (null-safe: DeleteObject on a
    // null HGDIOBJ fails harmlessly) and the icon copy — each exactly once.
    unsafe {
        let _ = DeleteObject(ii.hbmColor.into());
        let _ = DeleteObject(ii.hbmMask.into());
        let _ = DestroyIcon(icon);
    }
    out.map(|(rgba, w, h)| {
        let hot_x = ii.xHotspot.min(w.saturating_sub(1));
        let hot_y = ii.yHotspot.min(h.saturating_sub(1));
        (rgba, w, h, hot_x, hot_y)
    })
 }
 type RasterOut = Option<(Vec<u8>, u32, u32, u32, u32)>;
 /// Convert the ICONINFO bitmaps to straight RGBA. Two families:
 /// - color (`hbmColor` set): 32bpp BGRA; if the alpha channel is entirely empty (old-style
 ///   cursors) the AND mask supplies it (mask bit 1 = transparent).
 /// - monochrome (`hbmColor` null): `hbmMask` is DOUBLE height — AND plane over XOR plane, the
 ///   WebRTC truth table: (0,0) black, (0,1) white, (1,0) transparent, (1,1) invert. Invert
 ///   pixels — unrepresentable in straight alpha — become opaque black with a white outline
 ///   grown into adjacent transparency (the WebRTC approximation; keeps the I-beam legible on
 ///   any background, which the old translucent-gray stand-in did not).
 fn convert(ii: &ICONINFO) -> Option<(Vec<u8>, u32, u32)> {
    // SAFETY: GetDC(None) yields the screen DC, released below on every path; it is only used
    // as the GetDIBits reference DC.
    let dc = unsafe { GetDC(None) };
    let result = (|| {
        if !ii.hbmColor.is_invalid() {
            let color = read_bitmap_32(dc, ii.hbmColor)?;
            let (w, h) = (color.w as u32, color.h as u32);
            let mut rgba = bgra_to_rgba(&color.bgra);
            if rgba.chunks_exact(4).all(|p| p[3] == 0) {
                // Alpha-less color cursor: transparency lives in the AND mask.
                let mask = read_bitmap_32(dc, ii.hbmMask)?;
                if mask.w != color.w || mask.h < color.h {
                    return None;
                }
                for (px, m) in rgba.chunks_exact_mut(4).zip(mask.bgra.chunks_exact(4)) {
                    px[3] = if m[0] != 0 { 0 } else { 0xFF }; // mask white (AND=1) = transparent
                }
            }
            Some((rgba, w, h))
        } else {
            let mask = read_bitmap_32(dc, ii.hbmMask)?;
            if mask.h < 2 || mask.h % 2 != 0 {
                return None;
            }
            let (w, h) = (mask.w as usize, (mask.h / 2) as usize);
            let row = w * 4;
            let (and_plane, xor_plane) = mask.bgra.split_at(h * row);
            let mut rgba = vec![0u8; w * h * 4];
            let mut invert = vec![false; w * h];
            for i in 0..w * h {
                let (a, x) = (and_plane[i * 4] != 0, xor_plane[i * 4] != 0);
                let px = &mut rgba[i * 4..i * 4 + 4];
                match (a, x) {
                    (false, false) => px.copy_from_slice(&[0, 0, 0, 0xFF]),
                    (false, true) => px.copy_from_slice(&[0xFF, 0xFF, 0xFF, 0xFF]),
                    (true, false) => {} // transparent (already zeroed)
                    (true, true) => {
                        px.copy_from_slice(&[0, 0, 0, 0xFF]);
                        invert[i] = true;
                    }
                }
            }
            // White outline around invert regions so the (now black) shape survives dark
            // backgrounds: any transparent 8-neighbor of an invert pixel turns opaque white.
            for y in 0..h as i32 {
                for x in 0..w as i32 {
                    if !invert[(y * w as i32 + x) as usize] {
                        continue;
                    }
                    for (dx, dy) in NEIGHBORS {
                        let (nx, ny) = (x + dx, y + dy);
                        if nx < 0 || ny < 0 || nx >= w as i32 || ny >= h as i32 {
                            continue;
                        }
                        let o = (ny * w as i32 + nx) as usize * 4;
                        if rgba[o + 3] == 0 {
                            rgba[o..o + 4].copy_from_slice(&[0xFF, 0xFF, 0xFF, 0xFF]);
                        }
                    }
                }
            }
            Some((rgba, w as u32, h as u32))
        }
    })();
    // SAFETY: releasing the screen DC obtained above, exactly once.
    unsafe {
        ReleaseDC(None, dc);
    }
    result
 }
 const NEIGHBORS: [(i32, i32); 8] = [
    (-1, -1),
    (0, -1),
    (1, -1),
    (-1, 0),
    (1, 0),
    (-1, 1),
    (0, 1),
    (1, 1),
 ];
 struct RawBitmap {
    w: i32,
    h: i32,
    /// 32bpp top-down BGRA rows, `w*h*4` (monochrome sources arrive expanded: 0x00/0xFF channels).
    bgra: Vec<u8>,
 }
 /// Read any GDI bitmap as 32bpp top-down via `GetDIBits` (which performs the 1bpp→32bpp
 /// expansion for the mask planes).
 fn read_bitmap_32(dc: HDC, hbm: HBITMAP) -> Option<RawBitmap> {
    let mut bm = BITMAP::default();
    // SAFETY: `bm` is a live out-param sized exactly as passed; GetObjectW only writes into it.
    let n = unsafe {
        GetObjectW(
            hbm.into(),
            std::mem::size_of::<BITMAP>() as i32,
            Some((&mut bm as *mut BITMAP).cast()),
        )
    };
    if n == 0 || bm.bmWidth <= 0 || bm.bmHeight <= 0 || bm.bmWidth > 512 || bm.bmHeight > 1024 {
        return None; // 512/1024: sanity caps (256² is the wire max; XL accessibility ≤ that)
    }
    let (w, h) = (bm.bmWidth, bm.bmHeight);
    let mut info = BITMAPINFO {
        bmiHeader: BITMAPINFOHEADER {
            biSize: std::mem::size_of::<BITMAPINFOHEADER>() as u32,
            biWidth: w,
            biHeight: -h, // top-down
            biPlanes: 1,
            biBitCount: 32,
            biCompression: BI_RGB.0,
            ..Default::default()
        },
        ..Default::default()
    };
    let mut buf = vec![0u8; (w as usize) * (h as usize) * 4];
    // SAFETY: `buf` spans exactly `h` rows of `w` 32bpp pixels as described by `info`; both are
    // live locals for this synchronous call, `hbm` is a live bitmap not selected into any DC
    // (fresh GetIconInfo copies).
    let rows = unsafe {
        GetDIBits(
            dc,
            hbm,
            0,
            h as u32,
            Some(buf.as_mut_ptr().cast()),
            &mut info,
            DIB_RGB_COLORS,
        )
    };
    (rows != 0).then_some(RawBitmap { w, h, bgra: buf })
 }
 fn bgra_to_rgba(bgra: &[u8]) -> Vec<u8> {
    let mut out = bgra.to_vec();
    for px in out.chunks_exact_mut(4) {
        px.swap(0, 2);
    }
    out
 }
@@ -876,10 +876,16 @@ impl Worker {
            );
            return;
        };
-        let pref = self
+        let pref = match self.pad_info(id) {
-            .pad_info(id)
+            // Steam Input's virtual pad standing in front of the Deck's built-in controls (the
-            .map(|p| p.pref)
+            // only-pad-forwarded case, [`Self::forwarded_ids`]): declare the DECK kind, not the
-            .unwrap_or(GamepadPref::Xbox360);
+            // wrapper's Xbox 360 identity. [`Self::auto_pref`] already resolves the SESSION
            // default this way, but a current host honors the per-pad arrival over the session
            // default — so without this the host builds an X-Box 360 pad on a real Deck.
            Some(p) if p.steam_virtual && is_steam_deck() => GamepadPref::SteamDeck,
            Some(p) => p.pref,
            None => GamepadPref::Xbox360,
        };
        match self.subsystem.open(sdl3::sys::joystick::SDL_JoystickID(id)) {
            Ok(pad) => {
                let mut slot = Slot::new(id, index, pref, pad);
@@ -44,6 +44,13 @@ pub struct SessionParams {
    /// Share the clipboard with this host (the per-host `KnownHost::clipboard_sync`). The
    /// bridge additionally needs the host to advertise `HOST_CAP_CLIPBOARD`.
    pub clipboard: bool,
    /// Advertise `quic::CLIENT_CAP_CURSOR`: this embedder renders the host cursor locally
    /// (the presenter's cursor channel, design/remote-desktop-sweep.md M2), so the host may
    /// stop compositing the pointer into the video. Only set when the embedder actually
    /// draws it (the SDL presenter in desktop mouse mode) — a session that advertises it
    /// without rendering streams with NO visible cursor. The host answers `HOST_CAP_CURSOR`
    /// when its capture can forward (Linux portal, not gamescope/Windows).
    pub cursor_forward: bool,
    /// Video decoder preference (Settings; `PUNKTFUNK_DECODER` overrides — see
    /// `video::Decoder::new`).
    pub decoder: String,
@@ -255,6 +262,11 @@ fn pump(
        // This display's HDR volume → the host's virtual-display EDID. The env hatch wins so an
        // A/B run can pin an exact peak (PUNKTFUNK_CLIENT_PEAK_NITS=600).
        punktfunk_core::client::display_hdr_env_override().or(params.display_hdr),
        if params.cursor_forward {
            punktfunk_core::quic::CLIENT_CAP_CURSOR
        } else {
            0
        },
        params.launch.clone(),
        params.pin,
        Some(params.identity),
@@ -325,6 +325,11 @@ pub fn connect_reject_message(reason: punktfunk_core::reject::RejectReason) -> S
            "Client and host versions don't match — update both to the same release.".into()
        }
        R::Busy => "The host is busy with another session.".into(),
        R::SetupFailed => {
            "The host accepted the connection but couldn't start the stream — the host's log \
             (web console → Log) has the cause."
                .into()
        }
    }
 }
@@ -66,7 +66,31 @@ pub const fn interface_guid_fields() -> (u32, u16, u16, [u8; 8]) {
 /// too ([`MIN_DRIVER_PROTOCOL_VERSION`]) and simply falls back to the re-arrival resize against it;
 /// a v4 driver serving an older (v3-asserting) host fails that host's strict handshake — ship
 /// driver+host together, as ever.
-pub const PROTOCOL_VERSION: u32 = 4;
+/// v5: ADDITIVE — the IddCx HARDWARE CURSOR channel (remote-desktop-sweep M2c):
 /// [`control::AddRequest::hw_cursor`] (the former tail `_reserved` — same size, same offsets)
 /// asks the driver to declare a hardware cursor for this monitor (DWM then EXCLUDES the pointer
 /// from the desktop frame and delivers shape/position out-of-band), and
 /// [`control::IOCTL_SET_CURSOR_CHANNEL`] delivers the host-created [`cursor::CursorShm`] section
 /// the driver's cursor thread seqlock-publishes into. Nothing existing changed; the host gates
 /// the feature on the handshake-reported version (`>= 5`) and keeps composited-cursor behavior
 /// against older drivers.
 /// v6: ADDITIVE — [`control::IOCTL_SET_CURSOR_FORWARD`] (the mid-stream cursor-render flip,
 /// remote-desktop-sweep §8): the client's mouse-model flip (un)declares the hardware cursor on a
 /// LIVE monitor, so the capture mouse model gets DWM's composited pointer back (full fidelity)
 /// and the desktop model gets exclusion + forwarding. Nothing existing changed; against a v5
 /// driver the unknown IOCTL fails and the host logs + keeps the declared-at-ADD behavior.
 /// v6 tail ext (no bump, the `AddRequest` luminance-tail discipline):
 /// [`control::AddReply::cursor_excluded`] — the driver reports whether its ADAPTER already
 /// carries a hardware-cursor declare from an earlier session. A declare is IRREVOCABLE
 /// (remote-desktop-sweep §8.6, proven on-glass) and its exclusion reaches EVERY later monitor of
 /// the adapter, not just the declaring target (on-glass 2026-07-23: a declare on one target left
 /// a different client's fresh target cursor-less): DWM never composites the software cursor back
 /// into any of the adapter's frames until the adapter resets. The host uses the flag to
 /// self-composite the pointer (GDI poller + blend) in sessions that never negotiate the
 /// cursor channel — without it those sessions are silently cursor-less. Both skews degrade
 /// cleanly: an old driver writes only the 20-byte reply prefix (host reads `0` = unknown/clean),
 /// an old host retrieves a 20-byte buffer (driver writes just the prefix).
 pub const PROTOCOL_VERSION: u32 = 6;
 /// The OLDEST driver protocol this host still drives (v4 is additive over v3 — see the v4 note on
 /// [`PROTOCOL_VERSION`]): a v3 driver lacks only `IOCTL_UPDATE_MODES`, which the host gates on the
@@ -110,6 +134,22 @@ pub mod control {
    /// identity (saved per-monitor DPI) and the driver's swap-chain/stash machinery survive. A v3
    /// driver fails this unknown IOCTL → the host falls back to the re-arrival resize.
    pub const IOCTL_UPDATE_MODES: u32 = ctl_code(0x907);
    /// Deliver a monitor's hardware-cursor channel (v5): the handle VALUE of the unnamed
    /// [`cursor::CursorShm`](crate::cursor) file mapping the host duplicated into the WUDFHost
    /// (same delivery model as [`IOCTL_SET_FRAME_CHANNEL`], no event — the host polls the
    /// seqlock at its encode-tick pace). Sent once after ADD, only for a monitor whose
    /// [`AddRequest::hw_cursor`] was set. The driver maps it, calls
    /// `IddCxMonitorSetupHardwareCursor`, and starts its cursor-query thread. Input
    /// [`SetCursorChannelRequest`].
    pub const IOCTL_SET_CURSOR_CHANNEL: u32 = ctl_code(0x908);
    /// Flip a LIVE monitor's hardware-cursor declaration (v6, the mid-stream cursor-render
    /// flip): `enable = 1` re-declares (`IddCxMonitorSetupHardwareCursor` — DWM excludes the
    /// pointer, the query/shape machinery resumes), `enable = 0` un-declares (the driver stops
    /// re-declaring on mode commits and asks the OS to revert to the software cursor — DWM
    /// composites the pointer into the frame, the pre-channel behavior the capture mouse
    /// model wants). Only meaningful for a monitor whose cursor channel was delivered. Input
    /// [`SetCursorForwardRequest`].
    pub const IOCTL_SET_CURSOR_FORWARD: u32 = ctl_code(0x909);
    /// `IOCTL_ADD` input. A monotonic `session_id` keys the monitor (the host's refcount manager owns
    /// collision safety — no more SudoVDA's 16-byte GUID + pid-mangling). The driver advertises this
@@ -147,9 +187,13 @@ pub mod control {
        /// The client display's min luminance in MILLI-nits (0.001 cd/m² — the CTA min-luminance
        /// range lives well below 1 nit) → Desired Content Min Luminance. `0` = unknown.
        pub min_luminance_millinits: u32,
-        /// Pads the `u64`-aligned struct to a multiple of 8 (Pod forbids implicit tail padding);
+        /// Non-zero = declare an IddCx HARDWARE CURSOR for this monitor (v5, remote-desktop-sweep
-        /// free expansion room for the next appended field.
+        /// M2c): DWM stops compositing the pointer into the frame and the driver publishes
-        pub _reserved: u32,
+        /// shape/position into the [`cursor::CursorShm`](crate::cursor) section delivered by
        /// [`IOCTL_SET_CURSOR_CHANNEL`]. Byte-compatible with the old tail `_reserved` (offset 36):
        /// an un-upgraded driver ignores it (cursor stays composited — the host already gates on
        /// the handshake version, this is defense in depth), an un-upgraded host sends `0` (off).
        pub hw_cursor: u32,
    }
    /// [`AddRequest`]'s size before the client-HDR luminance tail — the prefix an un-upgraded
@@ -174,8 +218,22 @@ pub mod control {
        /// `DuplicateHandle`, then [`IOCTL_SET_FRAME_CHANNEL`]). Reported per-ADD, not per-open, so a
        /// WUDFHost restart between sessions can never leave the host duplicating into a dead process.
        pub wudf_pid: u32,
        /// Non-zero = the ADAPTER already carries an IRREVOCABLE hardware-cursor declare from an
        /// earlier session (remote-desktop-sweep §8.6; reach is adapter-wide, not per-target —
        /// on-glass 2026-07-23): DWM excludes the pointer from every frame on every monitor until
        /// the adapter resets, and a session without the cursor channel must
        /// self-composite (GDI poller + blend) or stream a cursor-less desktop. Appended after
        /// [`ADD_REPLY_LEGACY_SIZE`] under the same dual-size discipline as the `AddRequest`
        /// luminance tail: an un-upgraded driver writes only the legacy prefix (the host's
        /// zero-initialized buffer then reads `0` = unknown/clean), an un-upgraded host retrieves a
        /// legacy-size buffer (the driver writes just the prefix).
        pub cursor_excluded: u32,
    }
    /// [`AddReply`]'s size before the `cursor_excluded` tail — the prefix an un-upgraded driver
    /// writes and an un-upgraded host retrieves (see the field docs).
    pub const ADD_REPLY_LEGACY_SIZE: usize = 20;
    /// `IOCTL_REMOVE` input.
    #[repr(C)]
    #[derive(Clone, Copy, Pod, Zeroable, Debug, PartialEq, Eq)]
@@ -253,6 +311,29 @@ pub mod control {
    /// at the compile-time maximum; `ring_len` says how many entries are live).
    pub const RING_LEN_USIZE: usize = RING_LEN as usize;
    /// `IOCTL_SET_CURSOR_CHANNEL` input (v5): the hardware-cursor section for one monitor.
    #[repr(C)]
    #[derive(Clone, Copy, Pod, Zeroable, Debug, PartialEq, Eq)]
    pub struct SetCursorChannelRequest {
        /// The OS target id from [`AddReply`] — which monitor this channel belongs to.
        pub target_id: u32,
        pub _pad: u32,
        /// The [`cursor::CursorShm`](crate::cursor) file-mapping handle VALUE, already duplicated
        /// into the driver's WUDFHost process ([`AddReply::wudf_pid`]).
        pub header_handle: u64,
    }
    /// `IOCTL_SET_CURSOR_FORWARD` input (v6): the mid-stream cursor-render flip for one monitor.
    #[repr(C)]
    #[derive(Clone, Copy, Pod, Zeroable, Debug, PartialEq, Eq)]
    pub struct SetCursorForwardRequest {
        /// The OS target id from [`AddReply`] — which monitor to flip.
        pub target_id: u32,
        /// `1` = declare the hardware cursor (exclude + forward), `0` = un-declare (DWM
        /// composites — the capture mouse model).
        pub enable: u32,
    }
    // Layout is load-bearing across the process boundary — pin it. (bytemuck's Pod derive already
    // rejects any internal padding; these assert the externally-visible sizes too.) The `offset_of!`
    // asserts additionally catch a SAME-SIZE field reorder, which the size+Pod checks alone miss.
@@ -269,13 +350,18 @@ pub mod control {
        assert!(offset_of!(AddRequest, max_luminance_nits) == ADD_REQUEST_LEGACY_SIZE);
        assert!(offset_of!(AddRequest, max_frame_avg_nits) == 28);
        assert!(offset_of!(AddRequest, min_luminance_millinits) == 32);
        // v5: the former tail `_reserved` — same offset, same total size (rename-only).
        assert!(offset_of!(AddRequest, hw_cursor) == 36);
        assert!(size_of::<AddRequest>() == 40);
-        assert!(size_of::<AddReply>() == 20);
+        assert!(size_of::<AddReply>() == 24);
        assert!(offset_of!(AddReply, adapter_luid_low) == 0);
        assert!(offset_of!(AddReply, adapter_luid_high) == 4);
        assert!(offset_of!(AddReply, target_id) == 8);
        assert!(offset_of!(AddReply, resolved_monitor_id) == 12);
        assert!(offset_of!(AddReply, wudf_pid) == 16);
        // The cursor-excluded tail starts exactly at the legacy boundary (prefix-compat).
        assert!(offset_of!(AddReply, cursor_excluded) == ADD_REPLY_LEGACY_SIZE);
        assert!(size_of::<SetFrameChannelRequest>() == 32 + 8 * RING_LEN_USIZE);
        assert!(offset_of!(SetFrameChannelRequest, target_id) == 0);
@@ -288,6 +374,13 @@ pub mod control {
        assert!(size_of::<RemoveRequest>() == 8);
        assert!(offset_of!(RemoveRequest, session_id) == 0);
        assert!(size_of::<SetCursorChannelRequest>() == 16);
        assert!(offset_of!(SetCursorChannelRequest, target_id) == 0);
        assert!(offset_of!(SetCursorChannelRequest, header_handle) == 8);
        assert!(size_of::<SetCursorForwardRequest>() == 8);
        assert!(offset_of!(SetCursorForwardRequest, target_id) == 0);
        assert!(offset_of!(SetCursorForwardRequest, enable) == 4);
        assert!(size_of::<UpdateModesRequest>() == 24);
        assert!(offset_of!(UpdateModesRequest, session_id) == 0);
        assert!(offset_of!(UpdateModesRequest, width) == 8);
@@ -960,6 +1053,84 @@ pub mod mouse {
    };
 }
 /// The v5 hardware-cursor channel (remote-desktop-sweep M2c): one unnamed file mapping per
 /// monitor, host-created, delivered by handle value ([`control::IOCTL_SET_CURSOR_CHANNEL`]).
 /// The DRIVER's cursor thread (woken by its IddCx `hNewCursorDataAvailable` event) seqlock-writes
 /// shape + position + visibility; the HOST reads at its encode-tick pace — no event crosses the
 /// boundary. Writer: bump [`CursorShm::seq`] to ODD, write fields (+ shape bytes when the OS said
 /// the shape changed), bump to EVEN. Reader: read seq (retry while odd), copy, re-read seq —
 /// unchanged ⇒ consistent snapshot. Position-only updates never touch the shape bytes, so a
 /// reader that skips unchanged `shape_id`s never copies torn pixels.
 pub mod cursor {
    use bytemuck::{Pod, Zeroable};
    /// First field of [`CursorShm`] — `b"PFCU"` little-endian; anything else = not attached yet.
    pub const CURSOR_MAGIC: u32 = u32::from_le_bytes(*b"PFCU");
    /// Max cursor side (px) the driver declares to the OS (`IDDCX_CURSOR_CAPS::MaxX/MaxY`) and
    /// the section's shape buffer is sized for. Windows XL accessibility cursors top out here;
    /// the host's wire forwarder downscales to its own cap anyway.
    pub const CURSOR_SHAPE_MAX: u32 = 256;
    /// Shape-buffer bytes: 32-bpp at the declared max.
    pub const CURSOR_SHAPE_BYTES: usize = (CURSOR_SHAPE_MAX * CURSOR_SHAPE_MAX * 4) as usize;
    /// Byte offset of the shape pixels inside the section (one cache-line-ish header).
    pub const CURSOR_SHAPE_OFFSET: usize = 64;
    /// Total section size.
    pub const CURSOR_SHM_SIZE: usize = CURSOR_SHAPE_OFFSET + CURSOR_SHAPE_BYTES;
    /// `IDDCX_CURSOR_SHAPE_TYPE` values mirrored for the host (the driver writes the OS value
    /// verbatim into [`CursorShm::cursor_type`]).
    pub const CURSOR_TYPE_MASKED_COLOR: u32 = 1;
    pub const CURSOR_TYPE_ALPHA: u32 = 2;
    /// The section header (the shape pixels follow at [`CURSOR_SHAPE_OFFSET`]). `x`/`y` are the
    /// shape's TOP-LEFT in desktop coordinates (the IddCx `IDARG_OUT_QUERY_HWCURSOR::X/Y`
    /// convention — position − hotspot, can be negative); `shape_id` is the OS's per-set counter
    /// (bumps on every shape set, the overlay serial); pixels are the OS's 32-bpp rows at
    /// `pitch` bytes (BGRA for ALPHA; color+mask for MASKED_COLOR — the host converts).
    #[repr(C)]
    #[derive(Clone, Copy, Pod, Zeroable, Debug, PartialEq, Eq)]
    pub struct CursorShm {
        pub magic: u32,
        /// Seqlock: odd = writer mid-update.
        pub seq: u32,
        pub visible: u32,
        pub cursor_type: u32,
        pub x: i32,
        pub y: i32,
        pub shape_id: u32,
        pub width: u32,
        pub height: u32,
        pub pitch: u32,
        pub hot_x: u32,
        pub hot_y: u32,
        /// Reserved expansion room up to [`CURSOR_SHAPE_OFFSET`].
        pub _reserved: [u32; 4],
    }
    // Layout is load-bearing across the process boundary — pin it.
    const _: () = {
        use core::mem::{offset_of, size_of};
        assert!(size_of::<CursorShm>() == 64);
        assert!(size_of::<CursorShm>() <= CURSOR_SHAPE_OFFSET);
        assert!(offset_of!(CursorShm, magic) == 0);
        assert!(offset_of!(CursorShm, seq) == 4);
        assert!(offset_of!(CursorShm, visible) == 8);
        assert!(offset_of!(CursorShm, cursor_type) == 12);
        assert!(offset_of!(CursorShm, x) == 16);
        assert!(offset_of!(CursorShm, y) == 20);
        assert!(offset_of!(CursorShm, shape_id) == 24);
        assert!(offset_of!(CursorShm, width) == 28);
        assert!(offset_of!(CursorShm, height) == 32);
        assert!(offset_of!(CursorShm, pitch) == 36);
        assert!(offset_of!(CursorShm, hot_x) == 40);
        assert!(offset_of!(CursorShm, hot_y) == 44);
    };
 }
 #[cfg(test)]
 mod tests {
    use super::*;
@@ -1079,7 +1250,7 @@ mod tests {
            max_luminance_nits: 800,
            max_frame_avg_nits: 400,
            min_luminance_millinits: 50, // 0.05 nits
-            _reserved: 0,
+            hw_cursor: 1,
        };
        let bytes = bytemuck::bytes_of(&req);
        assert_eq!(bytes.len(), 40);
@@ -1109,14 +1280,19 @@ mod tests {
            target_id: 262,
            resolved_monitor_id: 7,
            wudf_pid: 4242,
            cursor_excluded: 1,
        };
        let rbytes = bytemuck::bytes_of(&reply);
-        assert_eq!(rbytes.len(), 20);
+        assert_eq!(rbytes.len(), 24);
        assert_eq!(*bytemuck::from_bytes::<control::AddReply>(rbytes), reply);
        // resolved_monitor_id occupies the old `_reserved` slot at offset 12 — byte-compatible.
        assert_eq!(rbytes[12..16], 7u32.to_le_bytes());
        // The v2 duplication-target pid trails at offset 16.
        assert_eq!(rbytes[16..20], 4242u32.to_le_bytes());
        // The cursor-excluded tail rides after the legacy boundary; an un-upgraded driver writes
        // only the prefix, so a zero-filled tail reads as "unknown/clean" (see the field docs).
        assert_eq!(rbytes[20..24], 1u32.to_le_bytes());
        assert_eq!(control::ADD_REPLY_LEGACY_SIZE, 20);
    }
    #[test]
@@ -1161,11 +1337,39 @@ mod tests {
            req
        );
        assert_eq!(bytes[8..12], 2560u32.to_le_bytes());
-        // The compat window: v4 is additive over v3, so the host floor stays one below.
+        // The compat window: v4–v6 are additive over v3, so the host floor stays at 3.
-        assert_eq!(PROTOCOL_VERSION, 4);
+        assert_eq!(PROTOCOL_VERSION, 6);
        assert_eq!(MIN_DRIVER_PROTOCOL_VERSION, 3);
    }
    #[test]
    fn cursor_shm_layout_is_pinned() {
        use cursor::*;
        // The header must leave the shape offset intact whatever grows inside `_reserved`.
        assert_eq!(core::mem::size_of::<CursorShm>(), 64);
        assert_eq!(CURSOR_SHM_SIZE, 64 + 256 * 256 * 4);
        assert_eq!(CURSOR_MAGIC, u32::from_le_bytes(*b"PFCU"));
        // Seqlock snapshot discipline survives a bytemuck roundtrip.
        let hdr = CursorShm {
            magic: CURSOR_MAGIC,
            seq: 2,
            visible: 1,
            cursor_type: CURSOR_TYPE_ALPHA,
            x: -3,
            y: 7,
            shape_id: 42,
            width: 32,
            height: 32,
            pitch: 128,
            hot_x: 4,
            hot_y: 5,
            _reserved: [0; 4],
        };
        let bytes = bytemuck::bytes_of(&hdr);
        assert_eq!(*bytemuck::from_bytes::<CursorShm>(bytes), hdr);
        assert_eq!(bytes[16..20], (-3i32).to_le_bytes());
    }
    #[test]
    fn gamepad_names_and_magics_are_stable() {
        assert_eq!(gamepad::xusb_boot_name(0), "Global\\pfxusb-boot-0");
@@ -1,105 +0,0 @@
 // Cursor-overlay blend kernels for the CUDA/NVENC path (cursor-as-metadata). The cursor bitmap is
 // straight-alpha RGBA, row-packed (stride = curW*4). Blended into the encoder-OWNED NVENC input
 // surface — never the compositor's dmabuf. One thread per cursor pixel (ARGB / YUV444) or per 2x2
 // chroma block (NV12). Coefficients are BT.709 limited, matching rgb2yuv.comp so the cursor colour
 // matches the rest of the frame regardless of which zero-copy backend encodes it.
 //
 // Build (regenerate cursor_blend.ptx after editing):
 //   nvcc -ptx -arch=compute_75 cursor_blend.cu -o cursor_blend.ptx
 // PTX is JIT'd by the driver forward to the actual GPU, so a compute_75 (Turing) baseline runs on
 // every Turing-or-newer NVENC GPU. (CUDA 13's nvcc no longer targets pre-Turing archs.)
 typedef unsigned char u8;
 __device__ __forceinline__ u8 blend8(int dst, int src, int a) {
    return (u8)((src * a + dst * (255 - a)) / 255);
 }
 // Packed 4-byte surface. NVENC's ARGB format stores bytes B,G,R,A in memory; the cursor is R,G,B,A.
 extern "C" __global__ void blend_argb(
    u8* surf, int pitch, int surfW, int surfH,
    const u8* cur, int curW, int curH, int ox, int oy)
 {
    int cx = blockIdx.x * blockDim.x + threadIdx.x;
    int cy = blockIdx.y * blockDim.y + threadIdx.y;
    if (cx >= curW || cy >= curH) return;
    int px = ox + cx, py = oy + cy;
    if (px < 0 || py < 0 || px >= surfW || py >= surfH) return;
    const u8* s = cur + (size_t)(cy * curW + cx) * 4;
    int a = s[3];
    if (a == 0) return;
    u8* d = surf + (size_t)py * pitch + (size_t)px * 4;
    d[0] = blend8(d[0], s[2], a); // B <- cursor B
    d[1] = blend8(d[1], s[1], a); // G <- cursor G
    d[2] = blend8(d[2], s[0], a); // R <- cursor R
 }
 // Planar YUV444: three full-res planes stacked at base, base+plane, base+2*plane (plane=pitch*surfH).
 extern "C" __global__ void blend_yuv444(
    u8* base, int pitch, int surfW, int surfH,
    const u8* cur, int curW, int curH, int ox, int oy)
 {
    int cx = blockIdx.x * blockDim.x + threadIdx.x;
    int cy = blockIdx.y * blockDim.y + threadIdx.y;
    if (cx >= curW || cy >= curH) return;
    int px = ox + cx, py = oy + cy;
    if (px < 0 || py < 0 || px >= surfW || py >= surfH) return;
    const u8* s = cur + (size_t)(cy * curW + cx) * 4;
    int a = s[3];
    if (a == 0) return;
    float R = s[0], G = s[1], B = s[2];
    int Y = (int)(16.0f + 0.1826f * R + 0.6142f * G + 0.0620f * B + 0.5f);
    int U = (int)(128.0f - 0.1006f * R - 0.3386f * G + 0.4392f * B + 0.5f);
    int V = (int)(128.0f + 0.4392f * R - 0.3989f * G - 0.0403f * B + 0.5f);
    size_t plane = (size_t)pitch * surfH;
    u8* yp = base + (size_t)py * pitch + px;
    u8* up = base + plane + (size_t)py * pitch + px;
    u8* vp = base + 2 * plane + (size_t)py * pitch + px;
    *yp = blend8(*yp, Y, a);
    *up = blend8(*up, U, a);
    *vp = blend8(*vp, V, a);
 }
 // NV12: full-res Y plane + interleaved half-res UV plane. One thread per 2x2 luma block; each blends
 // up to four Y samples and one (alpha-weighted) UV sample.
 extern "C" __global__ void blend_nv12(
    u8* yb, int yPitch, u8* uvb, int uvPitch, int surfW, int surfH,
    const u8* cur, int curW, int curH, int ox, int oy)
 {
    int bx = blockIdx.x * blockDim.x + threadIdx.x;
    int by = blockIdx.y * blockDim.y + threadIdx.y;
    int base_cx = bx * 2, base_cy = by * 2;
    if (base_cx >= curW || base_cy >= curH) return;
    float ua = 0.0f, va = 0.0f, wa = 0.0f;
    int cnt = 0;
    for (int j = 0; j < 2; j++) {
        for (int i = 0; i < 2; i++) {
            int cx = base_cx + i, cy = base_cy + j;
            if (cx >= curW || cy >= curH) continue;
            int px = ox + cx, py = oy + cy;
            if (px < 0 || py < 0 || px >= surfW || py >= surfH) continue;
            const u8* s = cur + (size_t)(cy * curW + cx) * 4;
            int a = s[3];
            if (a == 0) continue;
            float R = s[0], G = s[1], B = s[2];
            int Y = (int)(16.0f + 0.1826f * R + 0.6142f * G + 0.0620f * B + 0.5f);
            u8* yp = yb + (size_t)py * yPitch + px;
            *yp = blend8(*yp, Y, a);
            ua += (128.0f - 0.1006f * R - 0.3386f * G + 0.4392f * B) * a;
            va += (128.0f + 0.4392f * R - 0.3989f * G - 0.0403f * B) * a;
            wa += a;
            cnt++;
        }
    }
    if (wa <= 0.0f || cnt == 0) return;
    // The chroma sample covering this block's top-left surface pixel.
    int uvx = (ox + base_cx) / 2;
    int uvy = (oy + base_cy) / 2;
    if (uvx < 0 || uvy < 0 || uvx * 2 >= surfW || uvy * 2 >= surfH) return;
    int U = (int)(ua / wa + 0.5f);
    int V = (int)(va / wa + 0.5f);
    int amean = (int)(wa / cnt + 0.5f);
    u8* uv = uvb + (size_t)uvy * uvPitch + (size_t)uvx * 2;
    uv[0] = blend8(uv[0], U, amean);
    uv[1] = blend8(uv[1], V, amean);
 }
@@ -1,576 +0,0 @@
 //
 // Generated by NVIDIA NVVM Compiler
 //
 // Compiler Build ID: CL-38244171
 // Cuda compilation tools, release 13.3, V13.3.73
 // Based on NVVM 7.0.1
 //
 .version 9.3
 .target sm_75
 .address_size 64
 	// .globl	blend_argb
 .visible .entry blend_argb(
 	.param .u64 blend_argb_param_0,
 	.param .u32 blend_argb_param_1,
 	.param .u32 blend_argb_param_2,
 	.param .u32 blend_argb_param_3,
 	.param .u64 blend_argb_param_4,
 	.param .u32 blend_argb_param_5,
 	.param .u32 blend_argb_param_6,
 	.param .u32 blend_argb_param_7,
 	.param .u32 blend_argb_param_8
 )
 {
 	.reg .pred 	%p<10>;
 	.reg .b16 	%rs<2>;
 	.reg .b32 	%r<34>;
 	.reg .b64 	%rd<17>;
 	ld.param.u64 	%rd2, [blend_argb_param_0];
 	ld.param.u32 	%r5, [blend_argb_param_1];
 	ld.param.u32 	%r6, [blend_argb_param_2];
 	ld.param.u32 	%r7, [blend_argb_param_3];
 	ld.param.u64 	%rd3, [blend_argb_param_4];
 	ld.param.u32 	%r8, [blend_argb_param_5];
 	ld.param.u32 	%r11, [blend_argb_param_6];
 	ld.param.u32 	%r9, [blend_argb_param_7];
 	ld.param.u32 	%r10, [blend_argb_param_8];
 	mov.u32 	%r12, %ntid.x;
 	mov.u32 	%r13, %ctaid.x;
 	mov.u32 	%r14, %tid.x;
 	mad.lo.s32 	%r1, %r13, %r12, %r14;
 	mov.u32 	%r15, %ntid.y;
 	mov.u32 	%r16, %ctaid.y;
 	mov.u32 	%r17, %tid.y;
 	mad.lo.s32 	%r2, %r16, %r15, %r17;
 	setp.ge.s32 	%p1, %r1, %r8;
 	setp.ge.s32 	%p2, %r2, %r11;
 	or.pred  	%p3, %p1, %p2;
 	@%p3 bra 	$L__BB0_4;
 	add.s32 	%r3, %r1, %r9;
 	add.s32 	%r4, %r2, %r10;
 	or.b32  	%r18, %r4, %r3;
 	setp.lt.s32 	%p4, %r18, 0;
 	setp.ge.s32 	%p5, %r3, %r6;
 	or.pred  	%p6, %p5, %p4;
 	setp.ge.s32 	%p7, %r4, %r7;
 	or.pred  	%p8, %p7, %p6;
 	@%p8 bra 	$L__BB0_4;
 	mad.lo.s32 	%r19, %r2, %r8, %r1;
 	mul.wide.s32 	%rd4, %r19, 4;
 	cvta.to.global.u64 	%rd5, %rd3;
 	add.s64 	%rd1, %rd5, %rd4;
 	ld.global.u8 	%rs1, [%rd1+3];
 	setp.eq.s16 	%p9, %rs1, 0;
 	@%p9 bra 	$L__BB0_4;
 	cvt.u32.u16 	%r20, %rs1;
 	mul.wide.s32 	%rd6, %r4, %r5;
 	mul.wide.s32 	%rd7, %r3, 4;
 	add.s64 	%rd8, %rd6, %rd7;
 	cvta.to.global.u64 	%rd9, %rd2;
 	add.s64 	%rd10, %rd9, %rd8;
 	ld.global.u8 	%r21, [%rd10];
 	ld.global.u8 	%r22, [%rd1+2];
 	xor.b32  	%r23, %r20, 255;
 	mul.lo.s32 	%r24, %r23, %r21;
 	mad.lo.s32 	%r25, %r22, %r20, %r24;
 	mul.wide.u32 	%rd11, %r25, -2139062143;
 	shr.u64 	%rd12, %rd11, 39;
 	st.global.u8 	[%rd10], %rd12;
 	ld.global.u8 	%r26, [%rd10+1];
 	ld.global.u8 	%r27, [%rd1+1];
 	mul.lo.s32 	%r28, %r23, %r26;
 	mad.lo.s32 	%r29, %r27, %r20, %r28;
 	mul.wide.u32 	%rd13, %r29, -2139062143;
 	shr.u64 	%rd14, %rd13, 39;
 	st.global.u8 	[%rd10+1], %rd14;
 	ld.global.u8 	%r30, [%rd10+2];
 	ld.global.u8 	%r31, [%rd1];
 	mul.lo.s32 	%r32, %r23, %r30;
 	mad.lo.s32 	%r33, %r31, %r20, %r32;
 	mul.wide.u32 	%rd15, %r33, -2139062143;
 	shr.u64 	%rd16, %rd15, 39;
 	st.global.u8 	[%rd10+2], %rd16;
 $L__BB0_4:
 	ret;
 }
 	// .globl	blend_yuv444
 .visible .entry blend_yuv444(
 	.param .u64 blend_yuv444_param_0,
 	.param .u32 blend_yuv444_param_1,
 	.param .u32 blend_yuv444_param_2,
 	.param .u32 blend_yuv444_param_3,
 	.param .u64 blend_yuv444_param_4,
 	.param .u32 blend_yuv444_param_5,
 	.param .u32 blend_yuv444_param_6,
 	.param .u32 blend_yuv444_param_7,
 	.param .u32 blend_yuv444_param_8
 )
 {
 	.reg .pred 	%p<10>;
 	.reg .b16 	%rs<5>;
 	.reg .f32 	%f<16>;
 	.reg .b32 	%r<49>;
 	.reg .b64 	%rd<14>;
 	ld.param.u64 	%rd2, [blend_yuv444_param_0];
 	ld.param.u32 	%r5, [blend_yuv444_param_1];
 	ld.param.u32 	%r6, [blend_yuv444_param_2];
 	ld.param.u32 	%r7, [blend_yuv444_param_3];
 	ld.param.u64 	%rd3, [blend_yuv444_param_4];
 	ld.param.u32 	%r8, [blend_yuv444_param_5];
 	ld.param.u32 	%r11, [blend_yuv444_param_6];
 	ld.param.u32 	%r9, [blend_yuv444_param_7];
 	ld.param.u32 	%r10, [blend_yuv444_param_8];
 	mov.u32 	%r12, %ntid.x;
 	mov.u32 	%r13, %ctaid.x;
 	mov.u32 	%r14, %tid.x;
 	mad.lo.s32 	%r1, %r13, %r12, %r14;
 	mov.u32 	%r15, %ntid.y;
 	mov.u32 	%r16, %ctaid.y;
 	mov.u32 	%r17, %tid.y;
 	mad.lo.s32 	%r2, %r16, %r15, %r17;
 	setp.ge.s32 	%p1, %r1, %r8;
 	setp.ge.s32 	%p2, %r2, %r11;
 	or.pred  	%p3, %p1, %p2;
 	@%p3 bra 	$L__BB1_4;
 	add.s32 	%r3, %r1, %r9;
 	add.s32 	%r4, %r2, %r10;
 	or.b32  	%r18, %r4, %r3;
 	setp.lt.s32 	%p4, %r18, 0;
 	setp.ge.s32 	%p5, %r3, %r6;
 	or.pred  	%p6, %p5, %p4;
 	setp.ge.s32 	%p7, %r4, %r7;
 	or.pred  	%p8, %p7, %p6;
 	@%p8 bra 	$L__BB1_4;
 	mad.lo.s32 	%r19, %r2, %r8, %r1;
 	mul.wide.s32 	%rd4, %r19, 4;
 	cvta.to.global.u64 	%rd5, %rd3;
 	add.s64 	%rd1, %rd5, %rd4;
 	ld.global.u8 	%rs1, [%rd1+3];
 	setp.eq.s16 	%p9, %rs1, 0;
 	@%p9 bra 	$L__BB1_4;
 	cvt.u32.u16 	%r20, %rs1;
 	ld.global.u8 	%rs2, [%rd1];
 	cvt.rn.f32.u16 	%f1, %rs2;
 	ld.global.u8 	%rs3, [%rd1+1];
 	cvt.rn.f32.u16 	%f2, %rs3;
 	ld.global.u8 	%rs4, [%rd1+2];
 	cvt.rn.f32.u16 	%f3, %rs4;
 	fma.rn.f32 	%f4, %f1, 0f3E3AFB7F, 0f41800000;
 	fma.rn.f32 	%f5, %f2, 0f3F1D3C36, %f4;
 	fma.rn.f32 	%f6, %f3, 0f3D7DF3B6, %f5;
 	add.f32 	%f7, %f6, 0f3F000000;
 	cvt.rzi.s32.f32 	%r21, %f7;
 	fma.rn.f32 	%f8, %f1, 0fBDCE075F, 0f43000000;
 	fma.rn.f32 	%f9, %f2, 0fBEAD5CFB, %f8;
 	fma.rn.f32 	%f10, %f3, 0f3EE0DED3, %f9;
 	add.f32 	%f11, %f10, 0f3F000000;
 	cvt.rzi.s32.f32 	%r22, %f11;
 	fma.rn.f32 	%f12, %f1, 0f3EE0DED3, 0f43000000;
 	fma.rn.f32 	%f13, %f2, 0fBECC3C9F, %f12;
 	fma.rn.f32 	%f14, %f3, 0fBD25119D, %f13;
 	add.f32 	%f15, %f14, 0f3F000000;
 	cvt.rzi.s32.f32 	%r23, %f15;
 	mul.wide.s32 	%rd6, %r4, %r5;
 	cvt.s64.s32 	%rd7, %r3;
 	add.s64 	%rd8, %rd6, %rd7;
 	cvta.to.global.u64 	%rd9, %rd2;
 	add.s64 	%rd10, %rd9, %rd8;
 	ld.global.u8 	%r24, [%rd10];
 	mul.lo.s32 	%r25, %r21, %r20;
 	xor.b32  	%r26, %r20, 255;
 	mad.lo.s32 	%r27, %r26, %r24, %r25;
 	mul.hi.s32 	%r28, %r27, -2139062143;
 	add.s32 	%r29, %r28, %r27;
 	shr.u32 	%r30, %r29, 31;
 	shr.u32 	%r31, %r29, 7;
 	add.s32 	%r32, %r31, %r30;
 	st.global.u8 	[%rd10], %r32;
 	mul.wide.s32 	%rd11, %r7, %r5;
 	add.s64 	%rd12, %rd10, %rd11;
 	ld.global.u8 	%r33, [%rd12];
 	mul.lo.s32 	%r34, %r22, %r20;
 	mad.lo.s32 	%r35, %r26, %r33, %r34;
 	mul.hi.s32 	%r36, %r35, -2139062143;
 	add.s32 	%r37, %r36, %r35;
 	shr.u32 	%r38, %r37, 31;
 	shr.u32 	%r39, %r37, 7;
 	add.s32 	%r40, %r39, %r38;
 	st.global.u8 	[%rd12], %r40;
 	add.s64 	%rd13, %rd12, %rd11;
 	ld.global.u8 	%r41, [%rd13];
 	mul.lo.s32 	%r42, %r23, %r20;
 	mad.lo.s32 	%r43, %r26, %r41, %r42;
 	mul.hi.s32 	%r44, %r43, -2139062143;
 	add.s32 	%r45, %r44, %r43;
 	shr.u32 	%r46, %r45, 31;
 	shr.u32 	%r47, %r45, 7;
 	add.s32 	%r48, %r47, %r46;
 	st.global.u8 	[%rd13], %r48;
 $L__BB1_4:
 	ret;
 }
 	// .globl	blend_nv12
 .visible .entry blend_nv12(
 	.param .u64 blend_nv12_param_0,
 	.param .u32 blend_nv12_param_1,
 	.param .u64 blend_nv12_param_2,
 	.param .u32 blend_nv12_param_3,
 	.param .u32 blend_nv12_param_4,
 	.param .u32 blend_nv12_param_5,
 	.param .u64 blend_nv12_param_6,
 	.param .u32 blend_nv12_param_7,
 	.param .u32 blend_nv12_param_8,
 	.param .u32 blend_nv12_param_9,
 	.param .u32 blend_nv12_param_10
 )
 {
 	.reg .pred 	%p<43>;
 	.reg .b16 	%rs<17>;
 	.reg .f32 	%f<108>;
 	.reg .b32 	%r<123>;
 	.reg .b64 	%rd<35>;
 	ld.param.u64 	%rd11, [blend_nv12_param_0];
 	ld.param.u32 	%r21, [blend_nv12_param_1];
 	ld.param.u64 	%rd10, [blend_nv12_param_2];
 	ld.param.u32 	%r22, [blend_nv12_param_3];
 	ld.param.u32 	%r23, [blend_nv12_param_4];
 	ld.param.u32 	%r24, [blend_nv12_param_5];
 	ld.param.u64 	%rd12, [blend_nv12_param_6];
 	ld.param.u32 	%r25, [blend_nv12_param_7];
 	ld.param.u32 	%r26, [blend_nv12_param_8];
 	ld.param.u32 	%r27, [blend_nv12_param_9];
 	ld.param.u32 	%r28, [blend_nv12_param_10];
 	cvta.to.global.u64 	%rd1, %rd11;
 	cvta.to.global.u64 	%rd2, %rd12;
 	mov.u32 	%r29, %ntid.x;
 	mov.u32 	%r30, %ctaid.x;
 	mov.u32 	%r31, %tid.x;
 	mad.lo.s32 	%r32, %r30, %r29, %r31;
 	mov.u32 	%r33, %ntid.y;
 	mov.u32 	%r34, %ctaid.y;
 	mov.u32 	%r35, %tid.y;
 	mad.lo.s32 	%r36, %r34, %r33, %r35;
 	shl.b32 	%r1, %r32, 1;
 	shl.b32 	%r2, %r36, 1;
 	setp.ge.s32 	%p1, %r1, %r25;
 	setp.ge.s32 	%p2, %r2, %r26;
 	or.pred  	%p3, %p1, %p2;
 	mov.f32 	%f102, 0f00000000;
 	mov.f32 	%f103, 0f00000000;
 	mov.f32 	%f104, 0f00000000;
 	@%p3 bra 	$L__BB2_19;
 	cvt.s64.s32 	%rd3, %r21;
 	add.s32 	%r3, %r2, %r28;
 	setp.ge.s32 	%p4, %r3, %r24;
 	mul.lo.s32 	%r4, %r2, %r25;
 	mul.wide.s32 	%rd4, %r3, %r21;
 	add.s32 	%r5, %r1, %r27;
 	or.b32  	%r38, %r5, %r3;
 	setp.lt.s32 	%p5, %r38, 0;
 	mov.u32 	%r121, 0;
 	setp.ge.s32 	%p6, %r5, %r23;
 	or.pred  	%p7, %p6, %p5;
 	or.pred  	%p8, %p4, %p7;
 	@%p8 bra 	$L__BB2_4;
 	add.s32 	%r40, %r1, %r4;
 	mul.wide.s32 	%rd13, %r40, 4;
 	add.s64 	%rd5, %rd2, %rd13;
 	ld.global.u8 	%rs1, [%rd5+3];
 	setp.eq.s16 	%p9, %rs1, 0;
 	@%p9 bra 	$L__BB2_4;
 	cvt.u32.u16 	%r42, %rs1;
 	ld.global.u8 	%rs5, [%rd5];
 	cvt.rn.f32.u16 	%f31, %rs5;
 	ld.global.u8 	%rs6, [%rd5+1];
 	cvt.rn.f32.u16 	%f32, %rs6;
 	ld.global.u8 	%rs7, [%rd5+2];
 	cvt.rn.f32.u16 	%f33, %rs7;
 	fma.rn.f32 	%f34, %f31, 0f3E3AFB7F, 0f41800000;
 	fma.rn.f32 	%f35, %f32, 0f3F1D3C36, %f34;
 	fma.rn.f32 	%f36, %f33, 0f3D7DF3B6, %f35;
 	add.f32 	%f37, %f36, 0f3F000000;
 	cvt.rzi.s32.f32 	%r43, %f37;
 	cvt.s64.s32 	%rd14, %r5;
 	add.s64 	%rd15, %rd4, %rd14;
 	add.s64 	%rd16, %rd1, %rd15;
 	ld.global.u8 	%r44, [%rd16];
 	mul.lo.s32 	%r45, %r43, %r42;
 	xor.b32  	%r46, %r42, 255;
 	mad.lo.s32 	%r47, %r46, %r44, %r45;
 	mul.hi.s32 	%r48, %r47, -2139062143;
 	add.s32 	%r49, %r48, %r47;
 	shr.u32 	%r50, %r49, 31;
 	shr.u32 	%r51, %r49, 7;
 	add.s32 	%r52, %r51, %r50;
 	st.global.u8 	[%rd16], %r52;
 	fma.rn.f32 	%f38, %f31, 0fBDCE075F, 0f43000000;
 	fma.rn.f32 	%f39, %f32, 0fBEAD5CFB, %f38;
 	fma.rn.f32 	%f40, %f33, 0f3EE0DED3, %f39;
 	cvt.rn.f32.u16 	%f104, %rs1;
 	fma.rn.f32 	%f102, %f40, %f104, 0f00000000;
 	fma.rn.f32 	%f41, %f31, 0f3EE0DED3, 0f43000000;
 	fma.rn.f32 	%f42, %f32, 0fBECC3C9F, %f41;
 	fma.rn.f32 	%f43, %f33, 0fBD25119D, %f42;
 	fma.rn.f32 	%f103, %f43, %f104, 0f00000000;
 	mov.u32 	%r121, 1;
 $L__BB2_4:
 	add.s32 	%r7, %r1, 1;
 	setp.ge.s32 	%p10, %r7, %r25;
 	@%p10 bra 	$L__BB2_8;
 	add.s32 	%r8, %r7, %r27;
 	or.b32  	%r53, %r8, %r3;
 	setp.lt.s32 	%p12, %r53, 0;
 	setp.ge.s32 	%p13, %r8, %r23;
 	or.pred  	%p14, %p13, %p12;
 	or.pred  	%p15, %p4, %p14;
 	@%p15 bra 	$L__BB2_8;
 	add.s32 	%r54, %r7, %r4;
 	mul.wide.s32 	%rd17, %r54, 4;
 	add.s64 	%rd6, %rd2, %rd17;
 	ld.global.u8 	%rs2, [%rd6+3];
 	setp.eq.s16 	%p16, %rs2, 0;
 	@%p16 bra 	$L__BB2_8;
 	cvt.u32.u16 	%r55, %rs2;
 	ld.global.u8 	%rs8, [%rd6];
 	cvt.rn.f32.u16 	%f44, %rs8;
 	ld.global.u8 	%rs9, [%rd6+1];
 	cvt.rn.f32.u16 	%f45, %rs9;
 	ld.global.u8 	%rs10, [%rd6+2];
 	cvt.rn.f32.u16 	%f46, %rs10;
 	fma.rn.f32 	%f47, %f44, 0f3E3AFB7F, 0f41800000;
 	fma.rn.f32 	%f48, %f45, 0f3F1D3C36, %f47;
 	fma.rn.f32 	%f49, %f46, 0f3D7DF3B6, %f48;
 	add.f32 	%f50, %f49, 0f3F000000;
 	cvt.rzi.s32.f32 	%r56, %f50;
 	cvt.s64.s32 	%rd18, %r8;
 	add.s64 	%rd19, %rd4, %rd18;
 	add.s64 	%rd20, %rd1, %rd19;
 	ld.global.u8 	%r57, [%rd20];
 	mul.lo.s32 	%r58, %r56, %r55;
 	xor.b32  	%r59, %r55, 255;
 	mad.lo.s32 	%r60, %r59, %r57, %r58;
 	mul.hi.s32 	%r61, %r60, -2139062143;
 	add.s32 	%r62, %r61, %r60;
 	shr.u32 	%r63, %r62, 31;
 	shr.u32 	%r64, %r62, 7;
 	add.s32 	%r65, %r64, %r63;
 	st.global.u8 	[%rd20], %r65;
 	fma.rn.f32 	%f51, %f44, 0fBDCE075F, 0f43000000;
 	fma.rn.f32 	%f52, %f45, 0fBEAD5CFB, %f51;
 	fma.rn.f32 	%f53, %f46, 0f3EE0DED3, %f52;
 	cvt.rn.f32.u16 	%f54, %rs2;
 	fma.rn.f32 	%f102, %f53, %f54, %f102;
 	fma.rn.f32 	%f55, %f44, 0f3EE0DED3, 0f43000000;
 	fma.rn.f32 	%f56, %f45, 0fBECC3C9F, %f55;
 	fma.rn.f32 	%f57, %f46, 0fBD25119D, %f56;
 	fma.rn.f32 	%f103, %f57, %f54, %f103;
 	add.f32 	%f104, %f104, %f54;
 	add.s32 	%r121, %r121, 1;
 $L__BB2_8:
 	add.s32 	%r11, %r2, 1;
 	setp.ge.s32 	%p17, %r11, %r26;
 	add.s32 	%r12, %r11, %r28;
 	add.s32 	%r13, %r4, %r25;
 	cvt.s64.s32 	%rd21, %r12;
 	mul.lo.s64 	%rd7, %rd21, %rd3;
 	@%p17 bra 	$L__BB2_12;
 	setp.ge.s32 	%p18, %r12, %r24;
 	or.b32  	%r66, %r5, %r12;
 	setp.lt.s32 	%p19, %r66, 0;
 	or.pred  	%p21, %p6, %p19;
 	or.pred  	%p22, %p18, %p21;
 	@%p22 bra 	$L__BB2_12;
 	add.s32 	%r67, %r1, %r13;
 	mul.wide.s32 	%rd22, %r67, 4;
 	add.s64 	%rd8, %rd2, %rd22;
 	ld.global.u8 	%rs3, [%rd8+3];
 	setp.eq.s16 	%p23, %rs3, 0;
 	@%p23 bra 	$L__BB2_12;
 	cvt.u32.u16 	%r68, %rs3;
 	ld.global.u8 	%rs11, [%rd8];
 	cvt.rn.f32.u16 	%f58, %rs11;
 	ld.global.u8 	%rs12, [%rd8+1];
 	cvt.rn.f32.u16 	%f59, %rs12;
 	ld.global.u8 	%rs13, [%rd8+2];
 	cvt.rn.f32.u16 	%f60, %rs13;
 	fma.rn.f32 	%f61, %f58, 0f3E3AFB7F, 0f41800000;
 	fma.rn.f32 	%f62, %f59, 0f3F1D3C36, %f61;
 	fma.rn.f32 	%f63, %f60, 0f3D7DF3B6, %f62;
 	add.f32 	%f64, %f63, 0f3F000000;
 	cvt.rzi.s32.f32 	%r69, %f64;
 	cvt.s64.s32 	%rd23, %r5;
 	add.s64 	%rd24, %rd7, %rd23;
 	add.s64 	%rd25, %rd1, %rd24;
 	ld.global.u8 	%r70, [%rd25];
 	mul.lo.s32 	%r71, %r69, %r68;
 	xor.b32  	%r72, %r68, 255;
 	mad.lo.s32 	%r73, %r72, %r70, %r71;
 	mul.hi.s32 	%r74, %r73, -2139062143;
 	add.s32 	%r75, %r74, %r73;
 	shr.u32 	%r76, %r75, 31;
 	shr.u32 	%r77, %r75, 7;
 	add.s32 	%r78, %r77, %r76;
 	st.global.u8 	[%rd25], %r78;
 	fma.rn.f32 	%f65, %f58, 0fBDCE075F, 0f43000000;
 	fma.rn.f32 	%f66, %f59, 0fBEAD5CFB, %f65;
 	fma.rn.f32 	%f67, %f60, 0f3EE0DED3, %f66;
 	cvt.rn.f32.u16 	%f68, %rs3;
 	fma.rn.f32 	%f102, %f67, %f68, %f102;
 	fma.rn.f32 	%f69, %f58, 0f3EE0DED3, 0f43000000;
 	fma.rn.f32 	%f70, %f59, 0fBECC3C9F, %f69;
 	fma.rn.f32 	%f71, %f60, 0fBD25119D, %f70;
 	fma.rn.f32 	%f103, %f71, %f68, %f103;
 	add.f32 	%f104, %f104, %f68;
 	add.s32 	%r121, %r121, 1;
 $L__BB2_12:
 	or.pred  	%p26, %p17, %p10;
 	@%p26 bra 	$L__BB2_16;
 	setp.ge.s32 	%p27, %r12, %r24;
 	add.s32 	%r16, %r7, %r27;
 	or.b32  	%r79, %r16, %r12;
 	setp.lt.s32 	%p28, %r79, 0;
 	setp.ge.s32 	%p29, %r16, %r23;
 	or.pred  	%p30, %p29, %p28;
 	or.pred  	%p31, %p27, %p30;
 	@%p31 bra 	$L__BB2_16;
 	add.s32 	%r80, %r7, %r13;
 	mul.wide.s32 	%rd26, %r80, 4;
 	add.s64 	%rd9, %rd2, %rd26;
 	ld.global.u8 	%rs4, [%rd9+3];
 	setp.eq.s16 	%p32, %rs4, 0;
 	@%p32 bra 	$L__BB2_16;
 	cvt.u32.u16 	%r81, %rs4;
 	ld.global.u8 	%rs14, [%rd9];
 	cvt.rn.f32.u16 	%f72, %rs14;
 	ld.global.u8 	%rs15, [%rd9+1];
 	cvt.rn.f32.u16 	%f73, %rs15;
 	ld.global.u8 	%rs16, [%rd9+2];
 	cvt.rn.f32.u16 	%f74, %rs16;
 	fma.rn.f32 	%f75, %f72, 0f3E3AFB7F, 0f41800000;
 	fma.rn.f32 	%f76, %f73, 0f3F1D3C36, %f75;
 	fma.rn.f32 	%f77, %f74, 0f3D7DF3B6, %f76;
 	add.f32 	%f78, %f77, 0f3F000000;
 	cvt.rzi.s32.f32 	%r82, %f78;
 	cvt.s64.s32 	%rd27, %r16;
 	add.s64 	%rd28, %rd7, %rd27;
 	add.s64 	%rd29, %rd1, %rd28;
 	ld.global.u8 	%r83, [%rd29];
 	mul.lo.s32 	%r84, %r82, %r81;
 	xor.b32  	%r85, %r81, 255;
 	mad.lo.s32 	%r86, %r85, %r83, %r84;
 	mul.hi.s32 	%r87, %r86, -2139062143;
 	add.s32 	%r88, %r87, %r86;
 	shr.u32 	%r89, %r88, 31;
 	shr.u32 	%r90, %r88, 7;
 	add.s32 	%r91, %r90, %r89;
 	st.global.u8 	[%rd29], %r91;
 	fma.rn.f32 	%f79, %f72, 0fBDCE075F, 0f43000000;
 	fma.rn.f32 	%f80, %f73, 0fBEAD5CFB, %f79;
 	fma.rn.f32 	%f81, %f74, 0f3EE0DED3, %f80;
 	cvt.rn.f32.u16 	%f82, %rs4;
 	fma.rn.f32 	%f102, %f81, %f82, %f102;
 	fma.rn.f32 	%f83, %f72, 0f3EE0DED3, 0f43000000;
 	fma.rn.f32 	%f84, %f73, 0fBECC3C9F, %f83;
 	fma.rn.f32 	%f85, %f74, 0fBD25119D, %f84;
 	fma.rn.f32 	%f103, %f85, %f82, %f103;
 	add.f32 	%f104, %f104, %f82;
 	add.s32 	%r121, %r121, 1;
 $L__BB2_16:
 	setp.eq.s32 	%p33, %r121, 0;
 	setp.le.f32 	%p34, %f104, 0f00000000;
 	or.pred  	%p35, %p34, %p33;
 	@%p35 bra 	$L__BB2_19;
 	shr.u32 	%r92, %r5, 31;
 	add.s32 	%r93, %r5, %r92;
 	shr.s32 	%r19, %r93, 1;
 	setp.lt.s32 	%p36, %r3, -1;
 	setp.lt.s32 	%p37, %r5, -1;
 	or.pred  	%p38, %p37, %p36;
 	and.b32  	%r94, %r93, -2;
 	setp.ge.s32 	%p39, %r94, %r23;
 	or.pred  	%p40, %p38, %p39;
 	shr.u32 	%r95, %r3, 31;
 	add.s32 	%r96, %r3, %r95;
 	shr.s32 	%r20, %r96, 1;
 	and.b32  	%r97, %r96, -2;
 	setp.ge.s32 	%p41, %r97, %r24;
 	or.pred  	%p42, %p40, %p41;
 	@%p42 bra 	$L__BB2_19;
 	div.rn.f32 	%f86, %f102, %f104;
 	add.f32 	%f87, %f86, 0f3F000000;
 	cvt.rzi.s32.f32 	%r98, %f87;
 	div.rn.f32 	%f88, %f103, %f104;
 	add.f32 	%f89, %f88, 0f3F000000;
 	cvt.rzi.s32.f32 	%r99, %f89;
 	cvt.rn.f32.s32 	%f90, %r121;
 	div.rn.f32 	%f91, %f104, %f90;
 	add.f32 	%f92, %f91, 0f3F000000;
 	cvt.rzi.s32.f32 	%r100, %f92;
 	mul.wide.s32 	%rd30, %r20, %r22;
 	mul.wide.s32 	%rd31, %r19, 2;
 	add.s64 	%rd32, %rd30, %rd31;
 	cvta.to.global.u64 	%rd33, %rd10;
 	add.s64 	%rd34, %rd33, %rd32;
 	ld.global.u8 	%r101, [%rd34];
 	mul.lo.s32 	%r102, %r100, %r98;
 	mov.u32 	%r103, 255;
 	sub.s32 	%r104, %r103, %r100;
 	mad.lo.s32 	%r105, %r104, %r101, %r102;
 	mul.hi.s32 	%r106, %r105, -2139062143;
 	add.s32 	%r107, %r106, %r105;
 	shr.u32 	%r108, %r107, 31;
 	shr.u32 	%r109, %r107, 7;
 	add.s32 	%r110, %r109, %r108;
 	st.global.u8 	[%rd34], %r110;
 	ld.global.u8 	%r111, [%rd34+1];
 	mul.lo.s32 	%r112, %r100, %r99;
 	mad.lo.s32 	%r113, %r104, %r111, %r112;
 	mul.hi.s32 	%r114, %r113, -2139062143;
 	add.s32 	%r115, %r114, %r113;
 	shr.u32 	%r116, %r115, 31;
 	shr.u32 	%r117, %r115, 7;
 	add.s32 	%r118, %r117, %r116;
 	st.global.u8 	[%rd34+1], %r118;
 $L__BB2_19:
 	ret;
 }
@@ -69,6 +69,7 @@ use super::{AuChunk, ChromaFormat, Codec, EncodedFrame, Encoder, EncoderCaps};
 use anyhow::{anyhow, bail, Context, Result};
 use pf_frame::{CapturedFrame, FramePayload};
 use pf_zerocopy::cuda::{self, InputSurface};
 use pf_zerocopy::vkslot::{SlotFormat, VkSlotBlend, VkSlotRef};
 use std::collections::VecDeque;
 use std::ffi::c_void;
 use std::ptr;
@@ -76,12 +77,6 @@ use std::sync::mpsc;
 use nvidia_video_codec_sdk::sys::nvEncodeAPI as nv;
 /// Prebuilt PTX for the cursor-overlay blend kernels (cursor-as-metadata). Source is
 /// `cursor_blend.cu` beside this file; regenerate with
 /// `nvcc -ptx -arch=compute_75 cursor_blend.cu -o cursor_blend.ptx` after editing. JIT'd by the
 /// driver, so it runs on any Turing-or-newer GPU.
 const CURSOR_PTX: &[u8] = include_bytes!("cursor_blend.ptx");
 // ---------------------------------------------------------------------------------------------
 // Runtime-loaded NVENC entry table (Linux). Same shape as the Windows backend's `EncodeApi`, minus
 // the async-event entry points (Windows-only). Resolved once from `libnvidia-encode.so.1` — the two
@@ -428,13 +423,54 @@ fn buffer_format(buf: &cuda::DeviceBuffer) -> nv::NV_ENC_BUFFER_FORMAT {
    }
 }
-/// One encoder-owned CUDA input surface + its NVENC registration. The surface is copied into each
+/// One encoder-owned input surface + its NVENC registration. The surface is copied into each
 /// use (device→device) and the registration is created once at session init, unregistered at teardown.
 struct RingSlot {
-    surface: InputSurface,
+    surface: SlotSurface,
    reg: nv::NV_ENC_REGISTERED_PTR,
 }
 /// The ring slot's backing allocation: Vulkan external memory CUDA-imported (the normal case —
 /// blendable by the SPIR-V cursor pass, see `vkslot.rs`) or a plain pitched CUDA allocation (the
 /// fallback when Vulkan bring-up fails: sessions still encode, composite mode just has no
 /// cursor). Both present the same `(ptr, pitch, height)` NVENC-registration vocabulary.
 enum SlotSurface {
    Cuda(InputSurface),
    /// Backing objects live in the encoder's [`VkSlotBlend`] (freed by its `free_slots`); the
    /// ref itself is Copy and carries the registered geometry.
    Vk(VkSlotRef),
 }
 /// The [`SlotFormat`] for an NVENC buffer format (the ring-build + blend vocabulary).
 fn slot_fmt_of(fmt: nv::NV_ENC_BUFFER_FORMAT) -> SlotFormat {
    match fmt {
        nv::NV_ENC_BUFFER_FORMAT::NV_ENC_BUFFER_FORMAT_YUV444 => SlotFormat::Yuv444,
        nv::NV_ENC_BUFFER_FORMAT::NV_ENC_BUFFER_FORMAT_NV12 => SlotFormat::Nv12,
        _ => SlotFormat::Argb,
    }
 }
 impl SlotSurface {
    fn ptr(&self) -> pf_zerocopy::cuda::CUdeviceptr {
        match self {
            SlotSurface::Cuda(s) => s.ptr,
            SlotSurface::Vk(r) => r.ptr,
        }
    }
    fn pitch(&self) -> usize {
        match self {
            SlotSurface::Cuda(s) => s.pitch,
            SlotSurface::Vk(r) => r.pitch,
        }
    }
    fn height(&self) -> u32 {
        match self {
            SlotSurface::Cuda(s) => s.height,
            SlotSurface::Vk(r) => r.height,
        }
    }
 }
 /// `doNotWait` sampling cadence inside [`Encoder::poll_chunk`] — the probe measured ~200 µs
 /// between slice completions on the 5070 Ti, so 50 µs keeps the added per-chunk delivery delay
 /// well under one slice time without hammering the driver.
@@ -530,16 +566,23 @@ pub struct NvencCudaEncoder {
    split_mode: u32,
    /// The last reference-frame range we invalidated — dedupes repeated RFI requests for one loss.
    last_rfi_range: Option<(i64, i64)>,
-    /// Cursor-as-metadata GPU blend (loaded lazily on the first frame that carries a cursor, once the
+    /// Cursor-as-metadata GPU blend: the Vulkan device + SPIR-V compute pass the ring's
-    /// CUDA context is current). `None` until then or if the module load fails; `cursor_tried` stops
+    /// external-memory slots are allocated through (`vkslot.rs`) — the driver-portable
-    /// re-attempting a failed load every frame. `cursor_serial` tracks the uploaded bitmap.
+    /// replacement for the retired PTX kernels. Brought up once at session init (`cursor_tried`
-    cursor: Option<cuda::CursorBlend>,
+    /// stops re-attempts); `None` = bring-up failed, the ring fell back to plain CUDA
    /// allocations and composite mode degrades to no cursor. `cursor_serial` tracks the
    /// uploaded bitmap.
    vk_blend: Option<VkSlotBlend>,
    /// The session may hand this encoder cursor overlays (`SessionPlan.cursor_blend` — only
    /// cursor-channel sessions since Phase B). Off = skip the Vulkan bring-up entirely and ring
    /// on plain CUDA surfaces: embedded-pointer sessions never carry an overlay, so they pay
    /// zero blend cost, per-session or per-frame.
    blend_wanted: bool,
    cursor_tried: bool,
    cursor_serial: u64,
-    /// Suppress-until-success latches for the per-frame cursor upload/blend warns: a persistent
+    /// Suppress-until-success latch for the per-frame blend warn: a persistent failure sits in
-    /// failure sits in the submit() hot path, so warn once per failure streak (reset on success)
+    /// the submit() hot path, so warn once per failure streak (reset on success) rather than on
-    /// rather than on every cursor-bearing frame, which would evict the log ring.
+    /// every cursor-bearing frame, which would evict the log ring.
    cursor_upload_warned: bool,
    cursor_blend_warned: bool,
    /// One-shot latch for [`diagnose_failed_open`](Self::diagnose_failed_open) so a rebuild-retry
    /// burst (the session loop's bounded encoder resets) logs the diagnosis once, not per attempt.
@@ -607,6 +650,7 @@ impl NvencCudaEncoder {
        _cuda: bool,
        bit_depth: u8,
        chroma: ChromaFormat,
        cursor_blend: bool,
    ) -> Result<Self> {
        // The runtime `.so` load is the real "is NVENC possible here" gate: fail the open with a
        // clear reason instead of an opaque session error on the first frame.
@@ -645,10 +689,10 @@ impl NvencCudaEncoder {
            frame_idx: 0,
            force_kf: false,
            pending_anchor: false,
-            cursor: None,
+            vk_blend: None,
            blend_wanted: cursor_blend,
            cursor_tried: false,
            cursor_serial: u64::MAX,
            cursor_upload_warned: false,
            cursor_blend_warned: false,
            diagnosed: false,
            inited: false,
@@ -739,7 +783,12 @@ impl NvencCudaEncoder {
        // (the forfeit contract), and the next session re-latches the arming at init.
        self.subframe_chunks = false;
        self.chunk = None;
-        self.ring.clear(); // drops the InputSurfaces, freeing their CUDA allocations
+        self.ring.clear(); // drops the CUDA InputSurfaces; Vk slots are freed just below
        if let Some(vk) = &mut self.vk_blend {
            // The Vulkan-backed slots' memory (and its CUDA mapping) — the device itself stays
            // up for the next session's ring (`cursor_tried` keeps bring-up one-shot).
            vk.free_slots();
        }
        self.bitstreams.clear();
        self.pending.clear();
        self.encoder = ptr::null_mut();
@@ -1128,8 +1177,54 @@ impl NvencCudaEncoder {
            // Encoder-owned input-surface ring: allocate + register POOL surfaces in the negotiated
            // format. Registered once here, mapped per submit, unregistered at teardown.
            // Preferred backing = Vulkan external memory CUDA-imported (`vkslot.rs`), so the
            // SPIR-V cursor blend can composite into the very bytes NVENC encodes; any bring-up
            // or per-slot failure falls back to plain pitched CUDA allocations (sessions always
            // encode — composite mode just loses the cursor, warned below).
            if !self.cursor_tried && self.blend_wanted {
                self.cursor_tried = true;
                match VkSlotBlend::new() {
                    Ok(v) => self.vk_blend = Some(v),
                    Err(e) => tracing::warn!(
                        error = %format!("{e:#}"),
                        "NVENC (Linux): Vulkan slot-blend bring-up failed — plain CUDA input \
                         surfaces, cursor compositing unavailable"
                    ),
                }
            }
            let slot_fmt = slot_fmt_of(self.buffer_fmt);
            // Two attempts: the full ring on Vulkan slots, else (any failure) the full ring on
            // plain CUDA — never a mixed ring (it would blend on some slots only: a flickering
            // cursor) and never a short one.
            'ring: for use_vk in [self.vk_blend.is_some(), false] {
                if !use_vk && self.vk_blend.is_some() {
                    // Second attempt: retire the Vulkan side wholesale first.
                    for s in self.ring.drain(..) {
                        let _ = (api().unregister_resource)(self.encoder, s.reg);
                    }
                    if let Some(vk) = &mut self.vk_blend {
                        vk.free_slots();
                    }
                    self.vk_blend = None;
                }
                for _ in 0..POOL {
-                let surface = match self.buffer_fmt {
+                    let surface = if use_vk {
                        let vk = self.vk_blend.as_mut().expect("use_vk implies Some");
                        match vk.alloc_slot(slot_fmt, self.width, self.height) {
                            Ok(r) => SlotSurface::Vk(r),
                            Err(e) => {
                                tracing::warn!(
                                    error = %format!("{e:#}"),
                                    "NVENC (Linux): Vulkan slot alloc failed — rebuilding the \
                                     ring on plain CUDA surfaces (cursor compositing \
                                     unavailable)"
                                );
                                continue 'ring;
                            }
                        }
                    } else {
                        SlotSurface::Cuda(
                            match self.buffer_fmt {
                                nv::NV_ENC_BUFFER_FORMAT::NV_ENC_BUFFER_FORMAT_YUV444 => {
                                    InputSurface::alloc_yuv444(self.width, self.height)
                                }
@@ -1138,27 +1233,47 @@ impl NvencCudaEncoder {
                                }
                                _ => InputSurface::alloc_rgb(self.width, self.height),
                            }
-                .context("alloc NVENC input surface")?;
+                            .context("alloc NVENC input surface")?,
                        )
                    };
                    let mut rr = nv::NV_ENC_REGISTER_RESOURCE {
                        version: nv::NV_ENC_REGISTER_RESOURCE_VER,
                        resourceType:
                            nv::NV_ENC_INPUT_RESOURCE_TYPE::NV_ENC_INPUT_RESOURCE_TYPE_CUDADEVICEPTR,
                        width: self.width,
                        height: self.height,
-                    pitch: surface.pitch as u32,
+                        pitch: surface.pitch() as u32,
-                    resourceToRegister: surface.ptr as *mut c_void,
+                        resourceToRegister: surface.ptr() as *mut c_void,
                        bufferFormat: self.buffer_fmt,
                        bufferUsage: nv::NV_ENC_BUFFER_USAGE::NV_ENC_INPUT_IMAGE,
                        ..Default::default()
                    };
-                (api().register_resource)(self.encoder, &mut rr)
+                    match (api().register_resource)(self.encoder, &mut rr).nv_ok() {
-                    .nv_ok()
+                        Ok(()) => {}
-                    .map_err(|e| nvenc_status::call_err("register_resource (CUDADEVICEPTR)", e))?;
+                        Err(e) if use_vk => {
                            // NVENC refusing the imported pointer is a Vulkan-side condition
                            // too — same wholesale fallback.
                            tracing::warn!(
                                error = ?e,
                                "NVENC (Linux): registering a Vulkan-imported slot failed — \
                                 rebuilding the ring on plain CUDA surfaces"
                            );
                            continue 'ring;
                        }
                        Err(e) => {
                            return Err(nvenc_status::call_err(
                                "register_resource (CUDADEVICEPTR)",
                                e,
                            ))
                        }
                    }
                    self.ring.push(RingSlot {
                        surface,
                        reg: rr.registeredResource,
                    });
                }
                break 'ring; // full ring built
            }
            self.inited = true;
            // Two-thread retrieve (T2.2): spawn the lock thread against the live session. No
@@ -1242,9 +1357,9 @@ impl NvencCudaEncoder {
    /// IO-stream binding (stream-ordered submit — see the gate in [`Encoder::submit`]).
    fn copy_into_slot(&self, buf: &cuda::DeviceBuffer, slot: usize, sync: bool) -> Result<()> {
        let s = &self.ring[slot].surface;
-        let base = s.ptr;
+        let base = s.ptr();
-        let pitch = s.pitch;
+        let pitch = s.pitch();
-        let hh = s.height as u64;
+        let hh = s.height() as u64;
        match self.buffer_fmt {
            nv::NV_ENC_BUFFER_FORMAT::NV_ENC_BUFFER_FORMAT_YUV444 => {
                if !buf.yuv444 {
@@ -1396,74 +1511,85 @@ impl Encoder for NvencCudaEncoder {
        // `async_rt` must be absent too: in two-thread mode the frame may be recycled right after
        // submit returns while the stream still holds its copy (belt-and-braces — an escalated
        // session was rebuilt without the binding, so `stream_ordered` is false there anyway).
-        let ordered = self.stream_ordered && self.async_rt.is_none() && self.pending.is_empty();
+        // Cursor-bearing frames additionally force the CPU-synced path: the Vulkan blend sits
        // between the CUDA copy and the encode, and its cross-API ordering is fence/CPU-
        // established, not stream-ordered. Frames without a cursor (games hide it; client-draws
        // sessions strip it) keep the stream-ordered fast path untouched.
        let ordered = self.stream_ordered
            && self.async_rt.is_none()
            && self.pending.is_empty()
            && captured.cursor.is_none();
        let t0 = std::time::Instant::now();
        // Copy the captured buffer into this slot's input surface before encoding it.
        self.copy_into_slot(buf, slot, !ordered)?;
        let t_copy = t0.elapsed();
-        // Cursor-as-metadata: blend the overlay into this slot's OWNED input surface (a tiny kernel
+        // Cursor-as-metadata: blend the overlay into this slot's OWNED input surface via the
-        // over the cursor's rect — never the compositor's dmabuf). The PTX module loads lazily on the
+        // SPIR-V compute pass (a dispatch over the cursor's rect — never the compositor's
-        // first cursor frame now that the CUDA context is current; any failure degrades to no cursor,
+        // dmabuf). Cursor-bearing frames forced `ordered = false` above, so the CUDA copy has
-        // never a dropped frame.
+        // completed before the Vulkan dispatch and the fence-waited dispatch completes before
        // the encode below — the cross-API ordering is CPU-established. Any failure degrades to
        // no cursor, never a dropped frame.
        if let Some(ov) = &captured.cursor {
-            if !self.cursor_tried {
+            if let (Some(vk), SlotSurface::Vk(vref)) =
-                self.cursor_tried = true;
+                (self.vk_blend.as_mut(), &self.ring[slot].surface)
-                match cuda::CursorBlend::new(CURSOR_PTX) {
+            {
                    Ok(cb) => self.cursor = Some(cb),
                    Err(e) => tracing::warn!(
                        error = %format!("{e:#}"),
                        "NVENC (Linux): cursor blend module load failed — cursor not composited"
                    ),
                }
            }
            if let Some(cb) = &self.cursor {
                if self.cursor_serial != ov.serial {
-                    match cb.upload(ov.rgba.as_slice(), ov.w, ov.h) {
+                    vk.upload_cursor(ov.rgba.as_slice(), ov.w, ov.h);
                        Ok(()) => {
                    self.cursor_serial = ov.serial;
                            self.cursor_upload_warned = false;
                }
-                        Err(e) => {
+                // surfW = content width; the blend derives plane strides from the slot's luma
-                            if !self.cursor_upload_warned {
+                // height. Cursor pixels past the content land in cropped padding rows — harmless.
-                                self.cursor_upload_warned = true;
+                let r = vk.blend_ref(
-                                tracing::warn!(
+                    vref,
-                                    error = %format!("{e:#}"),
+                    slot_fmt_of(self.buffer_fmt),
-                                    serial = ov.serial,
+                    self.width,
-                                    "NVENC (Linux): cursor upload failed — cursor not composited"
+                    ov.w,
                    ov.h,
                    ov.x,
                    ov.y,
                );
                            }
                        }
                    }
                }
                let s = &self.ring[slot].surface;
                // surfW = content width; surfH = the surface's allocated height (matches
                // `copy_into_slot`'s plane math). Cursor pixels past the content are in cropped
                // padding rows — harmless.
                let (w, h) = (self.width, s.height);
                let r = match self.buffer_fmt {
                    nv::NV_ENC_BUFFER_FORMAT::NV_ENC_BUFFER_FORMAT_YUV444 => {
                        cb.blend_yuv444(s.ptr, s.pitch, w, h, ov.w, ov.h, ov.x, ov.y, !ordered)
                    }
                    nv::NV_ENC_BUFFER_FORMAT::NV_ENC_BUFFER_FORMAT_NV12 => {
                        cb.blend_nv12(s.ptr, s.pitch, w, h, ov.w, ov.h, ov.x, ov.y, !ordered)
                    }
                    _ => cb.blend_argb(s.ptr, s.pitch, w, h, ov.w, ov.h, ov.x, ov.y, !ordered),
                };
                if let Err(e) = r {
                    if !self.cursor_blend_warned {
                        self.cursor_blend_warned = true;
                        tracing::warn!(
                            error = %format!("{e:#}"),
-                            "NVENC (Linux): cursor blend launch failed — cursor not composited"
+                            "NVENC (Linux): cursor blend dispatch failed — cursor not composited"
                        );
                    }
                } else {
                    self.cursor_blend_warned = false;
                    // TEMP KWin composite probe (DROP BEFORE MERGE): prove the blend dispatches
                    // and with what geometry.
                    {
                        use std::sync::atomic::{AtomicU64, Ordering as ProbeOrd};
                        static PROBE_BLEND: AtomicU64 = AtomicU64::new(0);
                        let n = PROBE_BLEND.fetch_add(1, ProbeOrd::Relaxed) + 1;
                        if n == 1 || n % 512 == 0 {
                            tracing::info!(
                                n,
                                fmt = ?self.buffer_fmt,
                                ov_x = ov.x,
                                ov_y = ov.y,
                                ov_w = ov.w,
                                ov_h = ov.h,
                                visible = ov.visible,
                                "cursor blend probe: vulkan dispatch submitted"
                            );
                        }
                    }
                }
            } else if !self.cursor_blend_warned {
                self.cursor_blend_warned = true;
                tracing::warn!(
                    blend_wanted = self.blend_wanted,
                    "NVENC (Linux): cursor overlay present but no Vulkan blend (bring-up failed, \
                     or a non-blend session unexpectedly carried an overlay) — cursor not \
                     composited"
                );
            }
        }
        let t_blend = t0.elapsed() - t_copy;
        let t_map: std::time::Duration;
@@ -1509,7 +1635,7 @@ impl Encoder for NvencCudaEncoder {
                version: nv::NV_ENC_PIC_PARAMS_VER,
                inputWidth: self.width,
                inputHeight: self.height,
-                inputPitch: self.ring[slot].surface.pitch as u32,
+                inputPitch: self.ring[slot].surface.pitch() as u32,
                inputBuffer: mp.mappedResource,
                bufferFmt: mp.mappedBufferFmt,
                outputBitstream: self.bitstreams[slot],
@@ -270,7 +270,7 @@ fn open_video_backend(
    chroma: ChromaFormat,
    cursor_blend: bool,
 ) -> Result<(Box<dyn Encoder>, &'static str)> {
-    let _ = cursor_blend; // consumed only by the Linux vulkan-encode arm below
+    let _ = cursor_blend; // consumed only by the Linux vulkan-encode + direct-NVENC arms below
    validate_dimensions(codec, width, height)?;
    // Refresh/fps must be positive and sane: fps feeds the encoder time_base (`Rational(1, fps)`)
    // and the pts→ns conversion (`pts * 1e9 / fps`), so 0 builds a 1/0 rational / divides by zero.
@@ -394,6 +394,7 @@ fn open_video_backend(
                cuda,
                bit_depth,
                chroma,
                cursor_blend,
            )
            .map(|e| (e, "nvenc"))
        };
@@ -729,7 +730,10 @@ fn open_nvenc_probed(
    cuda: bool,
    bit_depth: u8,
    chroma: ChromaFormat,
    cursor_blend: bool,
 ) -> Result<Box<dyn Encoder>> {
    #[cfg(not(feature = "nvenc"))]
    let _ = cursor_blend; // consumed by the direct-SDK arm below
                          // Direct-SDK NVENC (design/linux-direct-nvenc.md): the DEFAULT on NVIDIA, and only for a CUDA
                          // capture payload (it registers CUDADEVICEPTR inputs — a CPU/dmabuf frame can't feed it, so those
                          // keep the libav path; and `cuda` is false on AMD/Intel, so they stay on VAAPI). Set
@@ -751,6 +755,7 @@ fn open_nvenc_probed(
            cuda,
            bit_depth,
            chroma,
            cursor_blend,
        )?) as Box<dyn Encoder>);
    }
    // The silent-degrade trap: a build without `--features nvenc` compiles the direct-SDK
@@ -32,6 +32,13 @@ pub struct WinCaptureTarget {
    /// duplicates the sealed frame channel's handles INTO (`idd_push::ChannelBroker`). `0` = unknown
    /// (a pre-v2 pairing can't occur — the version handshake is hard — so this only guards misuse).
    pub wudf_pid: u32,
    /// The ADD reply flagged the ADAPTER as carrying an IRREVOCABLE IddCx hardware-cursor declare
    /// from an earlier session (remote-desktop-sweep §8.6; reach is adapter-wide, not per-target —
    /// on-glass 2026-07-23, a GameStream session's fresh target streamed cursor-less): DWM
    /// excludes the pointer from its frames until adapter reset, so a session WITHOUT the cursor
    /// channel must self-composite (the IDD-push capturer's forced-composite gate) or the
    /// streamed desktop has no cursor at all.
    pub cursor_excluded: bool,
 }
 /// The PyroWave (Windows) zero-copy sharing payload attached to a captured frame: the SECOND plane
@@ -152,6 +152,12 @@ pub struct OutputFormat {
    /// codec's backend, and the fallback family) would misread the two-plane buffer as packed
    /// RGB. Always `false` on Windows (the IDD-push capturer owns its own formats).
    pub nv12_native: bool,
    /// The session negotiated the cursor-forward channel (remote-desktop-sweep M2c): on Windows
    /// the IDD-push capturer creates + delivers the driver's hardware-cursor section, so DWM
    /// stops compositing the pointer into the frames and the capturer surfaces it via
    /// `Capturer::cursor()` instead. Ignored on Linux (the portal's `SPA_META_Cursor` already
    /// separates the pointer; the session plan's `cursor_blend` gate handles the rest).
    pub hw_cursor: bool,
 }
 impl OutputFormat {
@@ -169,6 +175,8 @@ impl OutputFormat {
            chroma_444: false,
            // GameStream never negotiates PyroWave (native punktfunk/1 only).
            pyrowave: false,
            // GameStream/spike sessions never negotiate the cursor channel.
            hw_cursor: false,
            // Conservative: the GameStream + spike paths don't resolve the codec here, and a
            // Moonlight client may negotiate H264 (whose VAAPI backend can't ingest NV12) — so
            // they never prefer the producer-native NV12 pod. The punktfunk/1 plane opts in via
@@ -196,6 +204,16 @@ pub struct CursorOverlay {
    pub rgba: std::sync::Arc<Vec<u8>>,
    /// Bumps whenever `rgba`/`w`/`h` change; stable across position-only moves.
    pub serial: u64,
    /// Hotspot (the pixel that IS the pointer position) within `w`×`h`. The blend paths ignore
    /// it (`x`/`y` are already hotspot-adjusted); the cursor-forward channel ships it to the
    /// client so a locally-drawn OS cursor points with the right pixel.
    pub hot_x: u32,
    pub hot_y: u32,
    /// Compositor-reported pointer visibility. `false` = an app on the host grabbed/hid the
    /// pointer — the cursor-forward channel turns that into the client's relative-mode hint
    /// (remote-desktop-sweep M3). The encode loop STRIPS invisible overlays before the frame
    /// reaches any blend path, so encoders may keep treating `Some` as "draw it".
    pub visible: bool,
 }
 /// A captured frame. [`format`](Self::format)/dimensions describe the pixels regardless of
@@ -61,5 +61,10 @@ windows = { version = "0.62", features = [
    "Win32_System_StationsAndDesktops",
    "Win32_System_Threading",
    "Win32_UI_Input_KeyboardAndMouse",
    # Synthetic pointer devices (PT_PEN / PT_TOUCH) — the pen/touch injectors. The device
    # create/destroy + POINTER_TYPE_INFO live under Controls in this metadata layout;
    # InjectSyntheticPointerInput + POINTER_*_INFO under Input_Pointer.
    "Win32_UI_Controls",
    "Win32_UI_Input_Pointer",
    "Win32_UI_WindowsAndMessaging",
 ] }
@@ -36,6 +36,12 @@ pub fn vk_to_evdev(vk: u8) -> Option<u16> {
        0x2D => Some(110), // VK_INSERT   -> KEY_INSERT
        0x2E => Some(111), // VK_DELETE   -> KEY_DELETE
        // --- Consumer/media keys (Android TV remotes, keyboard media rows) ---
        0xB0 => Some(163), // VK_MEDIA_NEXT_TRACK -> KEY_NEXTSONG
        0xB1 => Some(165), // VK_MEDIA_PREV_TRACK -> KEY_PREVIOUSSONG
        0xB2 => Some(166), // VK_MEDIA_STOP       -> KEY_STOPCD
        0xB3 => Some(164), // VK_MEDIA_PLAY_PAUSE -> KEY_PLAYPAUSE
        // --- Generic modifiers ---
        0x10 => Some(42), // VK_SHIFT   -> KEY_LEFTSHIFT
        0x11 => Some(29), // VK_CONTROL -> KEY_LEFTCTRL
@@ -424,6 +424,9 @@ impl InputInjector for KwinFakeInjector {
                self.fake.touch_up(event.code);
                self.fake.touch_frame();
            }
            // fake_input can only press host-layout keycodes — no committed-text path (the
            // HOST_CAP_TEXT_INPUT cap is not advertised on this backend).
            InputKind::TextInput => {}
            // Gamepads are injected through uinput, not the compositor.
            InputKind::GamepadState
            | InputKind::GamepadButton
@@ -109,7 +109,7 @@ async fn session_main(mut rx: UnboundedReceiver<InputEvent>, source: EiSource) {
    // Keep `_rd`/`_session` bound for the whole loop — dropping the portal session closes the
    // EIS connection. Bound the setup so a headless approval dialog (un-bypassed grant) can't
    // hang the worker forever.
-    let (_keepalive, context, mut events) = match tokio::time::timeout(
+    let (_keepalive, context, mut events, output_hint) = match tokio::time::timeout(
        Duration::from_secs(30),
        connect(source),
    )
@@ -130,6 +130,7 @@ async fn session_main(mut rx: UnboundedReceiver<InputEvent>, source: EiSource) {
    tracing::info!("libei: EIS connected — awaiting devices");
    let mut state = EiState::new();
    state.output_hint = output_hint;
    // Watchdog: a healthy EIS server adds + resumes an input device within a beat of the handshake.
    // If none has resumed by this deadline, the connection is dead-on-arrival (stale/half-ready
    // gamescope socket the handshake passed but no real server is behind) — exit so the next
@@ -177,20 +178,28 @@ type Connected = (
    Box<dyn Send>,
    ei::Context,
    reis::tokio::EiConvertEventStream,
    // The compositor's output size ("WxH" relay-file hint) — the scale target for absolute
    // coordinates when the EIS advertises only a degenerate region (gamescope). `None` for
    // the portal/Mutter paths, whose regions are real.
    Option<(u32, u32)>,
 );
 /// Reach an EIS server per `source` and run the EI sender handshake.
 async fn connect(source: EiSource) -> Result<Connected> {
-    let (keepalive, stream): (Box<dyn Send>, UnixStream) = match source {
+    let (keepalive, stream, output_hint): (Box<dyn Send>, UnixStream, Option<(u32, u32)>) =
        match source {
            EiSource::Portal => {
                let (rd, session, fd) = connect_portal().await?;
-            (Box::new((rd, session)), UnixStream::from(fd))
+                (Box::new((rd, session)), UnixStream::from(fd), None)
            }
            EiSource::MutterEis => {
                let (keepalive, fd) = connect_mutter().await?;
-            (keepalive, UnixStream::from(fd))
+                (keepalive, UnixStream::from(fd), None)
            }
            EiSource::SocketPathFile(file) => {
                let (stream, hint) = connect_socket_file(&file).await?;
                (Box::new(()), stream, hint)
            }
        EiSource::SocketPathFile(file) => (Box::new(()), connect_socket_file(&file).await?),
        };
    let context = ei::Context::new(stream).map_err(|e| anyhow!("reis EI context: {e}"))?;
    // Bound the handshake. `UnixStream::connect` to a socket *file* succeeds the moment the path
@@ -206,7 +215,7 @@ async fn connect(source: EiSource) -> Result<Connected> {
        anyhow!("EI handshake timed out (EIS server not responding — stale/half-ready socket?)")
    })?
    .map_err(|e| anyhow!("EI handshake: {e}"))?;
-    Ok((keepalive, context, events))
+    Ok((keepalive, context, events, output_hint))
 }
 /// Open a RemoteDesktop portal session (pointer + keyboard) and obtain the EIS socket fd.
@@ -294,8 +303,10 @@ async fn connect_mutter() -> Result<(Box<dyn Send>, std::os::fd::OwnedFd)> {
 /// Poll `file` for the EIS socket path (the gamescope backend relays `LIBEI_SOCKET` there once
 /// the nested app launches), then connect. A bare name is resolved against `XDG_RUNTIME_DIR`,
-/// mirroring libei's own `LIBEI_SOCKET` semantics.
+/// mirroring libei's own `LIBEI_SOCKET` semantics. Line 2 of the relay file, when present,
-async fn connect_socket_file(file: &std::path::Path) -> Result<UnixStream> {
+/// carries the compositor's output size as `WxH` — returned as the absolute-coordinate scale
 /// hint (gamescope's EIS region is degenerate, so the geometry can't come from the protocol).
 async fn connect_socket_file(file: &std::path::Path) -> Result<(UnixStream, Option<(u32, u32)>)> {
    // The relay file is rewritten each session with the CURRENT gamescope's `LIBEI_SOCKET`, and the
    // socket may not be `listen()`ing the instant its name appears — or the file may briefly still
    // hold a prior, now-dead session's name (the host-lifetime injector reconnecting between
@@ -319,7 +330,12 @@ async fn connect_socket_file(file: &std::path::Path) -> Result<UnixStream> {
            ));
        }
        if let Ok(s) = std::fs::read_to_string(file) {
-            let name = s.trim();
+            let mut file_lines = s.lines();
            let name = file_lines.next().unwrap_or("").trim();
            let hint = file_lines.next().and_then(|l| {
                let (w, h) = l.trim().split_once('x')?;
                Some((w.parse::<u32>().ok()?, h.parse::<u32>().ok()?))
            });
            if !name.is_empty() {
                let full = if name.starts_with('/') {
                    std::path::PathBuf::from(name)
@@ -334,7 +350,7 @@ async fn connect_socket_file(file: &std::path::Path) -> Result<UnixStream> {
                    logged = name.to_string();
                }
                match UnixStream::connect(&full) {
-                    Ok(stream) => return Ok(stream),
+                    Ok(stream) => return Ok((stream, hint)),
                    // Refused = socket file exists but no listener yet (or a dead session);
                    // NotFound = path not created yet. Both heal once the live gamescope's EIS is
                    // up — retry. Anything else (e.g. permission) is a real failure.
@@ -358,6 +374,25 @@ async fn connect_socket_file(file: &std::path::Path) -> Result<UnixStream> {
 }
 /// One EI device and its emulation state.
 /// Pick the region to map absolute coordinates into: the one whose logical size matches the
 /// streamed mode (the session's virtual output). The device advertises one region per logical
 /// monitor, and blindly taking `first()` next to a physical monitor put the pointer — and every
 /// click — on whichever output the compositor happened to announce first (on-glass: GNOME with a
 /// dummy HDMI beside the virtual primary; the seat cursor never entered the streamed monitor, so
 /// neither embedded nor metadata cursor capture could see it). Size is the only key available
 /// today: regions carry no output name, and matching the screencast stream's `mapping_id` needs
 /// the stream id plumbed across crates (follow-up). Two same-sized monitors stay ambiguous.
 fn region_for_mode(
    regions: &[reis::event::Region],
    w: f32,
    h: f32,
 ) -> Option<&reis::event::Region> {
    regions
        .iter()
        .find(|r| r.width as f32 == w && r.height as f32 == h)
        .or_else(|| regions.first())
 }
 struct DeviceSlot {
    device: reis::event::Device,
    /// The device is resumed (allowed to emit). Devices arrive paused and may pause again.
@@ -386,6 +421,22 @@ struct EiState {
    held_keys: Vec<u32>,
    held_buttons: Vec<u32>,
    held_touches: Vec<u32>,
    /// The touch id currently degraded to the absolute pointer ([`EiState::degrade_touch`]) —
    /// the "primary finger" while the EIS has no touchscreen device. `None` between touches.
    degraded_touch: Option<u32>,
    /// The compositor's output size (relay-file "WxH" hint) — scale target for absolute
    /// coordinates when the device's region is degenerate/absent (gamescope). Without it the
    /// fallback is raw client pixels, correct only when the stream runs at the output's size.
    output_hint: Option<(u32, u32)>,
 }
 /// Is this EIS region a plausible OUTPUT geometry — something to map normalized coordinates
 /// into? gamescope advertises a degenerate `(0,0,INT32_MAX,INT32_MAX)` "everything" region on
 /// its virtual input device, meaning "absolute coordinates are raw"; normalizing into it
 /// explodes a center tap to x≈1e9, which the compositor clamps to the far corner. 16384
 /// comfortably covers real multi-monitor layouts while rejecting the sentinel.
 fn sane_region(r: &reis::event::Region) -> bool {
    r.width > 0 && r.height > 0 && r.width <= 16_384 && r.height <= 16_384
 }
 /// Stable small index per [`InputKind`] for the `seen_kinds` bitmask.
@@ -406,6 +457,7 @@ fn kind_bit(kind: InputKind) -> u32 {
        InputKind::GamepadState => 12,
        InputKind::GamepadRemove => 13,
        InputKind::GamepadArrival => 14,
        InputKind::TextInput => 15,
    };
    1 << i
 }
@@ -422,6 +474,8 @@ impl EiState {
            held_keys: Vec::new(),
            held_buttons: Vec::new(),
            held_touches: Vec::new(),
            degraded_touch: None,
            output_hint: None,
        }
    }
@@ -430,6 +484,10 @@ impl EiState {
    /// normal [`EiState::inject`] path so the compositor sees proper key-up / button-up /
    /// touch-up frames before the devices disappear.
    fn release_all(&mut self, ctx: &ei::Context) {
        // A degraded touch is held as a synthesized left button (in `held_buttons`), so the
        // loop below releases it — but the primary-finger latch must clear too, or the NEXT
        // session's first TouchDown reads as a second finger and is ignored.
        self.degraded_touch = None;
        let (keys, buttons, touches) = (
            std::mem::take(&mut self.held_keys),
            std::mem::take(&mut self.held_buttons),
@@ -506,6 +564,14 @@ impl EiState {
                    keyboard = dev.has_capability(DeviceCapability::Keyboard),
                    button = dev.has_capability(DeviceCapability::Button),
                    scroll = dev.has_capability(DeviceCapability::Scroll),
                    // One region per logical monitor — which one absolute coords map into is
                    // resolved per event (`region_for_mode`); log them so a mis-mapped pointer
                    // (cursor/clicks on the wrong output) is diagnosable from the journal.
                    regions = ?dev
                        .regions()
                        .iter()
                        .map(|r| format!("{}x{}+{}+{}", r.width, r.height, r.x, r.y))
                        .collect::<Vec<_>>(),
                    "libei: device RESUMED (now emittable)"
                );
            }
@@ -537,8 +603,85 @@ impl EiState {
        }
    }
    /// Degrade touch to a single-finger ABSOLUTE POINTER on compositors whose EIS never
    /// creates a touchscreen device (gamescope's "Gamescope Virtual Input" advertises
    /// pointer/pointer_abs/button but no touch — observed live; headless KWin the same).
    /// Down = abs-move + left press, move = abs-move, up = left release — synthesized through
    /// the normal [`EiState::inject`] Mouse* paths so region mapping, held-state tracking and
    /// [`EiState::release_all`] all apply. Only the FIRST finger drives the pointer; later
    /// fingers are ignored (a pointer has no second contact — a pinch degrades to a drag).
    fn degrade_touch(&mut self, ev: &InputEvent, ctx: &ei::Context) {
        const GS_BUTTON_LEFT: u32 = 1;
        match ev.kind {
            InputKind::TouchDown => {
                if self.degraded_touch.is_some() {
                    return; // secondary finger — single-pointer degradation
                }
                self.degraded_touch = Some(ev.code);
                static NOTED: std::sync::atomic::AtomicBool =
                    std::sync::atomic::AtomicBool::new(false);
                if !NOTED.swap(true, std::sync::atomic::Ordering::Relaxed) {
                    tracing::info!(
                        "compositor's EIS has no touchscreen device — degrading touch to a \
                         single-finger absolute pointer (tap = left click; multi-touch \
                         gestures unavailable)"
                    );
                }
                self.inject(
                    &InputEvent {
                        kind: InputKind::MouseMoveAbs,
                        ..*ev
                    },
                    ctx,
                );
                self.inject(
                    &InputEvent {
                        kind: InputKind::MouseButtonDown,
                        code: GS_BUTTON_LEFT,
                        ..*ev
                    },
                    ctx,
                );
            }
            InputKind::TouchMove if self.degraded_touch == Some(ev.code) => {
                self.inject(
                    &InputEvent {
                        kind: InputKind::MouseMoveAbs,
                        ..*ev
                    },
                    ctx,
                );
            }
            InputKind::TouchUp if self.degraded_touch == Some(ev.code) => {
                self.degraded_touch = None;
                self.inject(
                    &InputEvent {
                        kind: InputKind::MouseButtonUp,
                        code: GS_BUTTON_LEFT,
                        ..*ev
                    },
                    ctx,
                );
            }
            _ => {}
        }
    }
    /// Translate and emit one client input event, committing it as a single `frame`.
    fn inject(&mut self, ev: &InputEvent, ctx: &ei::Context) {
        // No ei_touchscreen device but an absolute pointer exists → degrade rather than drop
        // (the game-mode "touch simply does not work" trap). Checked per event, not latched:
        // a touchscreen device appearing later (compositor restart, capability change) takes
        // over seamlessly on the next touch.
        if matches!(
            ev.kind,
            InputKind::TouchDown | InputKind::TouchMove | InputKind::TouchUp
        ) && self.device_for(DeviceCapability::Touch).is_none()
            && self.device_for(DeviceCapability::PointerAbsolute).is_some()
        {
            self.degrade_touch(ev, ctx);
            return;
        }
        let cap = match ev.kind {
            InputKind::MouseMove => DeviceCapability::Pointer,
            InputKind::MouseMoveAbs => DeviceCapability::PointerAbsolute,
@@ -553,6 +696,9 @@ impl EiState {
            | InputKind::GamepadAxis
            | InputKind::GamepadRemove
            | InputKind::GamepadArrival => return, // uinput path (later)
            // libei presses keycodes against the server's negotiated keymap — no committed-text
            // path (the HOST_CAP_TEXT_INPUT cap is not advertised on this backend).
            InputKind::TextInput => return,
        };
        self.injected += 1;
        let n = self.injected;
@@ -605,16 +751,33 @@ impl EiState {
            InputKind::MouseMoveAbs => {
                let w = ((ev.flags >> 16) & 0xffff) as f32;
                let h = (ev.flags & 0xffff) as f32;
-                match (
+                match slot.interface::<ei::PointerAbsolute>() {
-                    slot.interface::<ei::PointerAbsolute>(),
+                    Some(p) if w > 0.0 && h > 0.0 => {
-                    slot.regions().first(),
+                        // Map the normalized client position into the region matching the streamed
-                ) {
+                        // output's mode (`region_for_mode` picks the right one on a multi-monitor
-                    (Some(p), Some(region)) if w > 0.0 && h > 0.0 => {
+                        // EIS). gamescope's "Gamescope Virtual Input" advertises a degenerate
-                        // Map the normalized client position into the device's first region.
+                        // (0,0,INT32_MAX,INT32_MAX) region meaning "coordinates are raw" —
                        // `sane_region` rejects it (normalizing into it explodes a center tap to
                        // x≈1e9, clamped to the far corner), so a non-matching / insane region
                        // falls to the output hint (correct across a resolution mismatch), then
                        // raw client pixels as the last resort.
                        let nx = (ev.x as f32 / w).clamp(0.0, 1.0);
                        let ny = (ev.y as f32 / h).clamp(0.0, 1.0);
-                        let x = region.x as f32 + nx * region.width as f32;
+                        let (x, y) = match region_for_mode(slot.regions(), w, h)
-                        let y = region.y as f32 + ny * region.height as f32;
+                            .filter(|r| sane_region(r))
                        {
                            Some(region) => (
                                region.x as f32 + nx * region.width as f32,
                                region.y as f32 + ny * region.height as f32,
                            ),
                            // Degenerate/absent region: scale into the relay-file output hint
                            // (correct even when the client streams at a different resolution
                            // than the session runs); raw client pixels as the last resort.
                            None => match self.output_hint {
                                Some((ow, oh)) => (nx * ow as f32, ny * oh as f32),
                                None => (ev.x as f32, ev.y as f32),
                            },
                        };
                        p.motion_absolute(x, y);
                    }
                    _ => emitted = false,
@@ -680,12 +843,23 @@ impl EiState {
            InputKind::TouchDown | InputKind::TouchMove => {
                let w = ((ev.flags >> 16) & 0xffff) as f32;
                let h = (ev.flags & 0xffff) as f32;
-                match (slot.interface::<ei::Touchscreen>(), slot.regions().first()) {
+                match slot.interface::<ei::Touchscreen>() {
-                    (Some(t), Some(region)) if w > 0.0 && h > 0.0 => {
+                    Some(t) if w > 0.0 && h > 0.0 => {
                        let nx = (ev.x as f32 / w).clamp(0.0, 1.0);
                        let ny = (ev.y as f32 / h).clamp(0.0, 1.0);
-                        let x = region.x as f32 + nx * region.width as f32;
+                        // Same region-selection + degenerate fallback ladder as MouseMoveAbs.
-                        let y = region.y as f32 + ny * region.height as f32;
+                        let (x, y) = match region_for_mode(slot.regions(), w, h)
                            .filter(|r| sane_region(r))
                        {
                            Some(region) => (
                                region.x as f32 + nx * region.width as f32,
                                region.y as f32 + ny * region.height as f32,
                            ),
                            None => match self.output_hint {
                                Some((ow, oh)) => (nx * ow as f32, ny * oh as f32),
                                None => (ev.x as f32, ev.y as f32),
                            },
                        };
                        if ev.kind == InputKind::TouchDown {
                            t.down(ev.code, x, y);
                        } else {
@@ -703,7 +877,8 @@ impl EiState {
            | InputKind::GamepadButton
            | InputKind::GamepadAxis
            | InputKind::GamepadRemove
-            | InputKind::GamepadArrival => emitted = false,
+            | InputKind::GamepadArrival
            | InputKind::TextInput => emitted = false,
        }
        if emitted {
@@ -0,0 +1,396 @@
 //! Virtual tablet ("Punktfunk Pen"): a uinput stylus device carrying the pen plane's full
 //! fidelity — pressure, tilt, barrel roll, hover distance, eraser, barrel buttons
 //! (design/pen-tablet-input.md §5).
 //!
 //! Deliberately a **uinput device, not a compositor-protocol citizen**: no virtual-tablet
 //! protocol exists in EI, the RemoteDesktop portal, KWin `fake_input`, or wlroots — while
 //! every compositor consumes evdev tablets via libinput and forwards them to apps over
 //! `zwp_tablet_v2` with nothing to configure. udev's `input_id` builtin classifies the device
 //! from its capabilities (`BTN_TOOL_PEN` + `ABS_X/Y` ⇒ `ID_INPUT_TABLET`), so Krita/GIMP/
 //! Xournal++ see a real pen. Output mapping is the compositor's own tablet-mapping default
 //! (single output ⇒ correct; multi-monitor pinning is the documented follow-up, which is why
 //! the device carries a stable, distinctive identity to key rules on).
 //!
 //! The consumer feeds it [`PenTransition`]s straight from the core's
 //! [`PenTracker`](punktfunk_core::quic::PenTracker); this file only translates transitions to
 //! evdev events and groups them into SYN frames so proximity-enter carries its position in the
 //! same frame (libinput would otherwise report an entry at a stale point).
 //!
 //! ioctl numbers/struct layouts mirror `gamepad.rs` (verified against the same kernel
 //! generation); each backend file stays self-contained by convention.
 use anyhow::{bail, Result};
 use punktfunk_core::quic::{PenSample, PenTool, PenTransition, PEN_BARREL1, PEN_BARREL2};
 use std::os::fd::{AsRawFd, OwnedFd};
 // ioctls (x86_64).
 const UI_DEV_CREATE: libc::c_ulong = 0x5501;
 const UI_DEV_DESTROY: libc::c_ulong = 0x5502;
 const UI_DEV_SETUP: libc::c_ulong = 0x405c_5503;
 const UI_ABS_SETUP: libc::c_ulong = 0x401c_5504;
 const UI_SET_EVBIT: libc::c_ulong = 0x4004_5564;
 const UI_SET_KEYBIT: libc::c_ulong = 0x4004_5565;
 const UI_SET_PROPBIT: libc::c_ulong = 0x4004_556e;
 // input-event-codes.h subset.
 const EV_SYN: u16 = 0x00;
 const EV_KEY: u16 = 0x01;
 const EV_ABS: u16 = 0x03;
 const SYN_REPORT: u16 = 0;
 const ABS_X: u16 = 0x00;
 const ABS_Y: u16 = 0x01;
 /// Barrel roll rides ABS_Z (the Wacom Art-Pen rotation convention); libinput normalizes the
 /// declared min..max onto its 0..360° rotation axis.
 const ABS_Z: u16 = 0x02;
 const ABS_PRESSURE: u16 = 0x18;
 const ABS_DISTANCE: u16 = 0x19;
 const ABS_TILT_X: u16 = 0x1a;
 const ABS_TILT_Y: u16 = 0x1b;
 const BTN_TOOL_PEN: u16 = 0x140;
 const BTN_TOOL_RUBBER: u16 = 0x141;
 const BTN_TOUCH: u16 = 0x14a;
 const BTN_STYLUS: u16 = 0x14b;
 const BTN_STYLUS2: u16 = 0x14c;
 /// The pen writes on the display it is mapped to (a "screen tablet"), not a desk pad —
 /// libinput then maps the full ABS range onto the output rect, exactly the wire's normalized
 /// coordinate contract.
 const INPUT_PROP_DIRECT: libc::c_int = 0x01;
 /// Full-scale wire pressure (u16) → the declared 0..4095 axis.
 const PRESSURE_SHIFT: u32 = 4;
 /// Wire hover distance (u16, 0xFFFF = unknown) → the declared 0..1023 axis.
 const DISTANCE_SHIFT: u32 = 6;
 const ABS_RANGE: f32 = 65535.0;
 #[repr(C)]
 struct InputId {
    bustype: u16,
    vendor: u16,
    product: u16,
    version: u16,
 }
 #[repr(C)]
 struct UinputSetup {
    id: InputId,
    name: [u8; 80],
    ff_effects_max: u32,
 }
 #[repr(C)]
 #[derive(Default, Clone, Copy)]
 struct AbsInfo {
    value: i32,
    minimum: i32,
    maximum: i32,
    fuzz: i32,
    flat: i32,
    resolution: i32,
 }
 #[repr(C)]
 struct UinputAbsSetup {
    code: u16,
    _pad: u16,
    absinfo: AbsInfo,
 }
 #[repr(C)]
 #[derive(Clone, Copy)]
 struct InputEventRaw {
    time: libc::timeval,
    type_: u16,
    code: u16,
    value: i32,
 }
 fn ioctl_int(fd: i32, req: libc::c_ulong, arg: libc::c_int, what: &str) -> Result<()> {
    // SAFETY: every caller passes a UI_SET_*/UI_DEV_* request whose argument the kernel reads
    // as a plain int; `fd` is a live uinput fd owned by the caller. No memory is handed over.
    if unsafe { libc::ioctl(fd, req, arg) } < 0 {
        bail!("{what}: {}", std::io::Error::last_os_error());
    }
    Ok(())
 }
 fn ioctl_ptr<T>(fd: i32, req: libc::c_ulong, arg: *mut T, what: &str) -> Result<()> {
    // SAFETY: every caller passes a pointer to a live, initialized `#[repr(C)]` struct matching
    // the request's expected layout (UI_DEV_SETUP/UI_ABS_SETUP); the kernel reads it during the
    // call and retains nothing.
    if unsafe { libc::ioctl(fd, req, arg) } < 0 {
        bail!("{what}: {}", std::io::Error::last_os_error());
    }
    Ok(())
 }
 /// The active tool's evdev key, one in proximity at a time (Wacom semantics — the core's
 /// [`PenTracker`](punktfunk_core::quic::PenTracker) already re-enters proximity on a tool
 /// switch, so this only tracks which key to release on `ProximityOut`).
 fn tool_key(tool: PenTool) -> u16 {
    match tool {
        PenTool::Eraser => BTN_TOOL_RUBBER,
        // Unknown = a newer client's future tool — nearest ink-capable behavior is the pen.
        PenTool::Pen | PenTool::Unknown => BTN_TOOL_PEN,
    }
 }
 /// One per-session virtual tablet. Created lazily on the first pen batch (a session that never
 /// draws never creates a device), destroyed with the session (Drop → `UI_DEV_DESTROY`).
 pub struct VirtualPen {
    fd: OwnedFd,
    /// The tool key currently held in proximity (release target for `ProximityOut`).
    tool: u16,
    /// Whether the current SYN frame already carries a Motion — the frame-split trigger for
    /// consecutive samples in one batch.
    frame_has_motion: bool,
    /// Whether the current SYN frame has any unflushed events.
    frame_dirty: bool,
 }
 impl VirtualPen {
    pub fn create() -> Result<VirtualPen> {
        use std::os::fd::FromRawFd;
        // SAFETY: `c"/dev/uinput"` is a 'static NUL-terminated C string literal; `open` reads it
        // as a path, returns a fresh fd (or -1) and retains nothing.
        let raw = unsafe {
            libc::open(
                c"/dev/uinput".as_ptr(),
                libc::O_RDWR | libc::O_NONBLOCK | libc::O_CLOEXEC,
            )
        };
        if raw < 0 {
            bail!(
                "open /dev/uinput: {} (install the udev rule granting the 'input' group access \
                 — see scripts/60-punktfunk.rules — and add the user to the 'input' group)",
                std::io::Error::last_os_error()
            );
        }
        // SAFETY: `raw >= 0` here, a freshly-opened fd owned nowhere else; `OwnedFd` becomes the
        // unique owner and closes it exactly once on drop.
        let fd = unsafe { OwnedFd::from_raw_fd(raw) };
        ioctl_int(raw, UI_SET_EVBIT, EV_KEY as i32, "UI_SET_EVBIT(EV_KEY)")?;
        ioctl_int(raw, UI_SET_EVBIT, EV_ABS as i32, "UI_SET_EVBIT(EV_ABS)")?;
        for key in [
            BTN_TOOL_PEN,
            BTN_TOOL_RUBBER,
            BTN_TOUCH,
            BTN_STYLUS,
            BTN_STYLUS2,
        ] {
            ioctl_int(raw, UI_SET_KEYBIT, key as i32, "UI_SET_KEYBIT")?;
        }
        ioctl_int(
            raw,
            UI_SET_PROPBIT,
            INPUT_PROP_DIRECT,
            "UI_SET_PROPBIT(DIRECT)",
        )?;
        // Position spans the full u16 range; `resolution` (units/mm) only feeds libinput's mm
        // math (nothing pen-relevant), but tablets without one trip its missing-resolution
        // fixup — 100 declares a plausible ~655 mm drawing surface.
        let pos = AbsInfo {
            minimum: 0,
            maximum: 65535,
            resolution: 100,
            ..Default::default()
        };
        // Tilt in degrees from vertical, per evdev convention; resolution = units/radian (57
        // ⇔ 1 unit = 1°, what the Wacom driver declares).
        let tilt = AbsInfo {
            minimum: -90,
            maximum: 90,
            resolution: 57,
            ..Default::default()
        };
        for (code, info) in [
            (ABS_X, pos),
            (ABS_Y, pos),
            (
                ABS_PRESSURE,
                AbsInfo {
                    minimum: 0,
                    maximum: 4095,
                    ..Default::default()
                },
            ),
            (
                ABS_DISTANCE,
                AbsInfo {
                    minimum: 0,
                    maximum: 1023,
                    ..Default::default()
                },
            ),
            (ABS_TILT_X, tilt),
            (ABS_TILT_Y, tilt),
            (
                // Barrel roll: libinput maps the declared range linearly onto 0..360°.
                ABS_Z,
                AbsInfo {
                    minimum: 0,
                    maximum: 359,
                    ..Default::default()
                },
            ),
        ] {
            let mut a = UinputAbsSetup {
                code,
                _pad: 0,
                absinfo: info,
            };
            ioctl_ptr(raw, UI_ABS_SETUP, &mut a, "UI_ABS_SETUP")?;
        }
        // A stable, distinctive identity (pid.codes open-source VID) so compositor
        // tablet-mapping rules — GNOME's per-`vendor:product` gsettings path, sway
        // `map_to_output` — can target exactly this device.
        let mut setup = UinputSetup {
            id: InputId {
                bustype: 0x0006, // BUS_VIRTUAL
                vendor: 0x1209,
                product: 0x5046, // "PF"
                version: 1,
            },
            name: [0; 80],
            ff_effects_max: 0,
        };
        let name = b"Punktfunk Pen";
        setup.name[..name.len()].copy_from_slice(name);
        ioctl_ptr(raw, UI_DEV_SETUP, &mut setup, "UI_DEV_SETUP")?;
        ioctl_int(raw, UI_DEV_CREATE, 0, "UI_DEV_CREATE")?;
        tracing::info!("virtual tablet created (Punktfunk Pen, uinput)");
        Ok(VirtualPen {
            fd,
            tool: BTN_TOOL_PEN,
            frame_has_motion: false,
            frame_dirty: false,
        })
    }
    fn emit(&self, type_: u16, code: u16, value: i32) {
        let ev = InputEventRaw {
            time: libc::timeval {
                tv_sec: 0,
                tv_usec: 0,
            },
            type_,
            code,
            value,
        };
        // SAFETY: `ev` is a live local `#[repr(C)]` all-integer struct (no padding: timeval=16 +
        // u16 + u16 + i32 = 24), so every byte is initialized; the slice spans exactly `ev`'s
        // bytes and is used immediately below with no concurrent mutation.
        let bytes = unsafe {
            std::slice::from_raw_parts(
                &ev as *const _ as *const u8,
                std::mem::size_of::<InputEventRaw>(),
            )
        };
        // Best-effort like the gamepad path: a full kernel queue drops the event; pen samples
        // are state-full, so the next frame re-syncs axes (and the tracker re-syncs state).
        // SAFETY: `self.fd` stays open for the synchronous call; `write` only reads
        // `bytes.len()` bytes from the still-live local and retains nothing.
        let _ = unsafe {
            libc::write(
                self.fd.as_raw_fd(),
                bytes.as_ptr() as *const libc::c_void,
                bytes.len(),
            )
        };
    }
    fn flush(&mut self) {
        if self.frame_dirty {
            self.emit(EV_SYN, SYN_REPORT, 0);
            self.frame_dirty = false;
            self.frame_has_motion = false;
        }
    }
    fn motion(&mut self, s: &PenSample) {
        self.emit(EV_ABS, ABS_X, (s.x * ABS_RANGE) as i32);
        self.emit(EV_ABS, ABS_Y, (s.y * ABS_RANGE) as i32);
        self.emit(EV_ABS, ABS_PRESSURE, (s.pressure >> PRESSURE_SHIFT) as i32);
        if s.distance != punktfunk_core::quic::PEN_DISTANCE_UNKNOWN {
            self.emit(EV_ABS, ABS_DISTANCE, (s.distance >> DISTANCE_SHIFT) as i32);
        }
        // Polar → tiltX/tiltY needs both angles; azimuth clockwise from north, so east (90°)
        // tilts +X and south (180°, toward the user) tilts +Y — the evdev/W3C signs.
        if s.tilt_deg != punktfunk_core::quic::PEN_TILT_UNKNOWN
            && s.azimuth_deg != punktfunk_core::quic::PEN_ANGLE_UNKNOWN
        {
            let az = (s.azimuth_deg as f32).to_radians();
            let tilt = s.tilt_deg as f32;
            self.emit(EV_ABS, ABS_TILT_X, (tilt * az.sin()).round() as i32);
            self.emit(EV_ABS, ABS_TILT_Y, (-tilt * az.cos()).round() as i32);
        }
        if s.roll_deg != punktfunk_core::quic::PEN_ANGLE_UNKNOWN {
            self.emit(EV_ABS, ABS_Z, (s.roll_deg % 360) as i32);
        }
        self.frame_dirty = true;
        self.frame_has_motion = true;
    }
    /// Apply one decoded batch's transitions (the core tracker's output, in its documented
    /// order), grouping them into SYN frames: a frame closes before a `ProximityIn` (an entry
    /// is a new instant — and must carry its own position, not inherit the stale frame) and
    /// before a second `Motion` (consecutive samples are consecutive instants), plus a final
    /// close. So `[ProxIn, Motion, TipDown]` lands as ONE frame — libinput reports the entry
    /// already at the right point with contact — while a drag batch's `[Motion, Motion]`
    /// stays two.
    pub fn apply_batch(&mut self, transitions: &[PenTransition]) {
        for t in transitions {
            match t {
                PenTransition::ProximityIn { tool } => {
                    self.flush();
                    self.tool = tool_key(*tool);
                    self.emit(EV_KEY, self.tool, 1);
                    self.frame_dirty = true;
                }
                PenTransition::Motion { sample } => {
                    if self.frame_has_motion {
                        self.flush();
                    }
                    self.motion(sample);
                }
                PenTransition::TipDown => {
                    self.emit(EV_KEY, BTN_TOUCH, 1);
                    self.frame_dirty = true;
                }
                PenTransition::ButtonsChanged { pressed, released } => {
                    for (bit, key) in [(PEN_BARREL1, BTN_STYLUS), (PEN_BARREL2, BTN_STYLUS2)] {
                        if pressed & bit != 0 {
                            self.emit(EV_KEY, key, 1);
                            self.frame_dirty = true;
                        }
                        if released & bit != 0 {
                            self.emit(EV_KEY, key, 0);
                            self.frame_dirty = true;
                        }
                    }
                }
                PenTransition::TipUp => {
                    self.emit(EV_KEY, BTN_TOUCH, 0);
                    self.emit(EV_ABS, ABS_PRESSURE, 0);
                    self.frame_dirty = true;
                }
                PenTransition::ProximityOut => {
                    self.emit(EV_KEY, self.tool, 0);
                    self.frame_dirty = true;
                }
            }
        }
        self.flush();
    }
 }
 impl Drop for VirtualPen {
    fn drop(&mut self) {
        // SAFETY: `self.fd` is still open (OwnedFd closes only after this body returns);
        // UI_DEV_DESTROY takes no pointer argument. Errors are moot on teardown.
        let _ = unsafe { libc::ioctl(self.fd.as_raw_fd(), UI_DEV_DESTROY, 0) };
    }
 }
@@ -98,9 +98,26 @@ pub struct WlrootsInjector {
    keyboard: ZwpVirtualKeyboardV1,
    xkb_state: xkb::State,
    _keymap_file: std::fs::File, // keep the memfd alive for the compositor's mmap
    /// Dedicated committed-text device ([`InputKind::TextInput`]), created on first use.
    text: Option<TextKeyboard>,
    start: Instant,
 }
 /// Cap on distinct characters the dynamic text keymap holds before it restarts from scratch
 /// (keycodes grow upward from 9; xkb tops out at 255, so stay well under).
 const TEXT_KEYMAP_MAX: usize = 200;
 /// The dedicated **text** virtual keyboard: types committed IME text (`InputKind::TextInput`,
 /// one Unicode scalar per event) by growing a keymap of Unicode keysyms on demand and pressing
 /// the character's keycode — the `wtype` model. A separate `zwp_virtual_keyboard` so keymap
 /// re-uploads never disturb the main device's layout/modifier state that VK key events ride on.
 struct TextKeyboard {
    keyboard: ZwpVirtualKeyboardV1,
    /// Characters in keycode order: `chars[i]` types on wire keycode `i + 1` (xkb `i + 9`).
    chars: Vec<char>,
    _keymap_file: Option<std::fs::File>, // keep the memfd alive for the compositor's mmap
 }
 impl WlrootsInjector {
    pub fn open() -> Result<Self> {
        let conn = Connection::connect_to_env()
@@ -171,6 +188,7 @@ impl WlrootsInjector {
            keyboard,
            xkb_state,
            _keymap_file: file,
            text: None,
            start: Instant::now(),
        })
    }
@@ -179,6 +197,54 @@ impl WlrootsInjector {
        self.start.elapsed().as_millis() as u32
    }
    /// Type one committed-text Unicode scalar on the dedicated text device (created lazily),
    /// growing its keymap when the character is new. Control characters are dropped — Enter,
    /// Backspace and Tab ride the VK key-event path.
    fn type_text(&mut self, cp: u32) -> Result<()> {
        let Some(ch) = char::from_u32(cp) else {
            return Ok(()); // lone surrogate / out of range
        };
        if ch.is_control() {
            return Ok(());
        }
        if self.text.is_none() {
            let (Some(mgr), Some(seat)) =
                (self.globals.keyboard_mgr.clone(), self.globals.seat.clone())
            else {
                return Ok(());
            };
            let kb = mgr.create_virtual_keyboard(&seat, &self.queue.handle(), ());
            self.text = Some(TextKeyboard {
                keyboard: kb,
                chars: Vec::new(),
                _keymap_file: None,
            });
        }
        let t = self.now_ms();
        let text = self.text.as_mut().expect("created above");
        let code = match text.chars.iter().position(|&c| c == ch) {
            Some(i) => (i + 1) as u32,
            None => {
                if text.chars.len() >= TEXT_KEYMAP_MAX {
                    text.chars.clear(); // restart the map; old codes are re-assigned lazily
                }
                text.chars.push(ch);
                let keymap_str = text_keymap(&text.chars);
                let file = memfd_with(&keymap_str)?;
                text.keyboard.keymap(
                    1, /* XKB_V1 */
                    file.as_fd(),
                    keymap_str.len() as u32 + 1,
                );
                text._keymap_file = Some(file);
                text.chars.len() as u32
            }
        };
        text.keyboard.key(t, code, 1);
        text.keyboard.key(t, code, 0);
        Ok(())
    }
    /// Update xkb state for a key and tell the compositor the resulting modifier mask.
    fn send_modifiers(&mut self, evdev: u16, down: bool) {
        let kc = xkb::Keycode::new(evdev as u32 + 8); // evdev -> xkb keycode
@@ -254,6 +320,9 @@ impl InputInjector for WlrootsInjector {
                    tracing::debug!(vk = event.code, "unmapped VK keycode — dropped");
                }
            }
            InputKind::TextInput => {
                self.type_text(event.code)?;
            }
            InputKind::GamepadState
            | InputKind::GamepadButton
            | InputKind::GamepadAxis
@@ -271,6 +340,33 @@ impl InputInjector for WlrootsInjector {
    }
 }
 /// Build a minimal xkb keymap whose keycode `i + 9` (wire code `i + 1`) types `chars[i]`, using
 /// Unicode keysym names (`U<hex>` — xkbcommon resolves them for any scalar, emoji included).
 /// Types/compat `include "complete"` mirrors `wtype`'s generated keymap — proven on wlroots
 /// compositors, and the system XKB data is present (the main keymap compiled from it in `open`).
 fn text_keymap(chars: &[char]) -> String {
    use std::fmt::Write as _;
    let mut keycodes = String::new();
    let mut symbols = String::new();
    for (i, ch) in chars.iter().enumerate() {
        let _ = writeln!(keycodes, "        <T{i}> = {};", i + 9);
        let _ = writeln!(symbols, "        key <T{i}> {{ [ U{:04X} ] }};", *ch as u32);
    }
    format!(
        "xkb_keymap {{\n\
             xkb_keycodes \"punktfunk-text\" {{\n\
                 minimum = 8;\n\
                 maximum = {};\n\
         {keycodes}\
             }};\n\
             xkb_types \"punktfunk-text\" {{ include \"complete\" }};\n\
             xkb_compatibility \"punktfunk-text\" {{ include \"complete\" }};\n\
             xkb_symbols \"punktfunk-text\" {{\n{symbols}    }};\n\
         }};\n",
        chars.len() + 9,
    )
 }
 /// Create an anonymous in-memory file holding `s` + a trailing NUL (for the keymap fd).
 fn memfd_with(s: &str) -> Result<std::fs::File> {
    let name = b"punktfunk-keymap\0";
@@ -0,0 +1,556 @@
 //! Windows synthetic-pointer injection (design/pen-tablet-input.md §6): a per-session `PT_PEN`
 //! device carrying the pen plane's full fidelity — pressure (rescaled to Windows' 0..1024),
 //! polar tilt → tiltX/tiltY, barrel roll on `rotation` (0..359 — Windows Ink renders Pencil
 //! Pro roll natively), barrel button, eraser (`PEN_FLAG_INVERTED`/`ERASER`), hover — plus a
 //! `PT_TOUCH` device that closes the long-standing SendInput touch no-op for wire touches.
 //!
 //! Both follow Apollo's proven recipe (`design/apollo-comparison.md`): synthetic pointer state
 //! goes STALE if not re-injected (~100 ms), so each device runs a small refresh thread that
 //! re-asserts the last frame every [`REFRESH_MS`] while the pen is in range / contacts are
 //! held — a stationary stylus must not hover-out (and a held finger must not auto-lift) just
 //! because no new samples arrived. `CreateSyntheticPointerDevice` needs Win10 1809+;
 //! [`crate::pen_supported`] probes it, so older hosts simply never advertise pen.
 //!
 //! Frame grouping mirrors the Linux uinput backend: a proximity-enter is injected together
 //! with its position (never at a stale point), tip edges get their own DOWN/UP frames, and a
 //! range-leave is a final frame without `INRANGE`.
 // Every `unsafe` block in this file carries a `// SAFETY:` proof; enforce it.
 #![deny(clippy::undocumented_unsafe_blocks)]
 use anyhow::{Context, Result};
 use punktfunk_core::input::{InputEvent, InputKind};
 use punktfunk_core::quic::{
    PenSample, PenTool, PenTransition, PEN_ANGLE_UNKNOWN, PEN_BARREL1, PEN_TILT_UNKNOWN,
 };
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::{Arc, Mutex};
 use windows::Win32::Foundation::POINT;
 use windows::Win32::UI::Controls::{
    CreateSyntheticPointerDevice, DestroySyntheticPointerDevice, HSYNTHETICPOINTERDEVICE,
    POINTER_FEEDBACK_DEFAULT, POINTER_TYPE_INFO, POINTER_TYPE_INFO_0,
 };
 use windows::Win32::UI::Input::Pointer::{
    InjectSyntheticPointerInput, POINTER_FLAGS, POINTER_FLAG_DOWN, POINTER_FLAG_FIRSTBUTTON,
    POINTER_FLAG_INCONTACT, POINTER_FLAG_INRANGE, POINTER_FLAG_NEW, POINTER_FLAG_UP,
    POINTER_FLAG_UPDATE, POINTER_INFO, POINTER_PEN_INFO, POINTER_TOUCH_INFO,
 };
 use windows::Win32::UI::WindowsAndMessaging::{
    GetSystemMetrics, PEN_FLAG_BARREL, PEN_FLAG_ERASER, PEN_FLAG_INVERTED, PEN_FLAG_NONE,
    PEN_MASK_PRESSURE, PEN_MASK_ROTATION, PEN_MASK_TILT_X, PEN_MASK_TILT_Y, PT_PEN, PT_TOUCH,
    SM_CXVIRTUALSCREEN, SM_CYVIRTUALSCREEN, SM_XVIRTUALSCREEN, SM_YVIRTUALSCREEN, TOUCH_FLAG_NONE,
    TOUCH_MASK_NONE,
 };
 /// Re-inject cadence while state is held (in-range pen / touching contacts). Apollo uses
 /// 50 ms against the ~100 ms staleness window; 40 ms keeps two refreshes inside it.
 const REFRESH_MS: u64 = 40;
 /// Windows pen pressure is 0..1024 (vs the wire's full-scale u16).
 const WIN_PEN_PRESSURE_MAX: u32 = 1024;
 /// Map a normalized [0,1] coordinate pair onto virtual-desktop pixels — the same surface the
 /// SendInput absolute mouse targets, so pen, touch, and pointer all land identically.
 fn to_screen(x: f32, y: f32) -> POINT {
    // SAFETY: `GetSystemMetrics` takes a constant index and reads global metrics; no memory in.
    let (vx, vy, vw, vh) = unsafe {
        (
            GetSystemMetrics(SM_XVIRTUALSCREEN),
            GetSystemMetrics(SM_YVIRTUALSCREEN),
            GetSystemMetrics(SM_CXVIRTUALSCREEN).max(1),
            GetSystemMetrics(SM_CYVIRTUALSCREEN).max(1),
        )
    };
    POINT {
        x: vx + (x.clamp(0.0, 1.0) * (vw - 1) as f32) as i32,
        y: vy + (y.clamp(0.0, 1.0) * (vh - 1) as f32) as i32,
    }
 }
 /// An owned `HSYNTHETICPOINTERDEVICE` (destroyed exactly once on drop).
 struct Device(HSYNTHETICPOINTERDEVICE);
 // SAFETY: the handle is a plain kernel object identifier with no thread affinity —
 // `InjectSyntheticPointerInput`/`DestroySyntheticPointerDevice` are documented callable from
 // any thread; ownership transfer/sharing does not alias memory.
 unsafe impl Send for Device {}
 impl Drop for Device {
    fn drop(&mut self) {
        // SAFETY: `self.0` is the device this wrapper uniquely owns; destroyed once here.
        unsafe { DestroySyntheticPointerDevice(self.0) };
    }
 }
 /// Probe: can this Windows build create a synthetic pen device (Win10 1809+)?
 pub fn synthetic_pen_available() -> bool {
    // SAFETY: FFI create with by-value args; on success the returned handle is destroyed
    // immediately by the `Device` wrapper, exactly once.
    match unsafe { CreateSyntheticPointerDevice(PT_PEN, 1, POINTER_FEEDBACK_DEFAULT) } {
        Ok(h) => {
            drop(Device(h));
            true
        }
        Err(_) => false,
    }
 }
 /// The tracked pen state a refresh tick re-asserts.
 #[derive(Default)]
 struct PenState {
    in_range: bool,
    touching: bool,
    barrel: bool,
    eraser: bool,
    x: f32,
    y: f32,
    pressure: u16,
    tilt_deg: u8,
    azimuth_deg: u16,
    roll_deg: u16,
 }
 struct PenShared {
    dev: Device,
    state: PenState,
    /// First injection failure logs at WARN (see `TouchShared::fail_warned`).
    fail_warned: bool,
 }
 /// One per-session virtual pen (the Windows sibling of the Linux uinput tablet — same
 /// [`PenTransition`] consumer API). Wire barrel button 2 has no Windows pen equivalent
 /// (one barrel + eraser is the platform model) and is ignored here.
 pub struct VirtualPen {
    shared: Arc<Mutex<PenShared>>,
    stop: Arc<AtomicBool>,
    refresher: Option<std::thread::JoinHandle<()>>,
    // Batch-local frame grouping (single-threaded within apply_batch).
    edge_down: bool,
    edge_up: bool,
    is_new: bool,
    frame_dirty: bool,
    frame_has_motion: bool,
 }
 impl VirtualPen {
    pub fn create() -> Result<VirtualPen> {
        // SAFETY: FFI create with by-value args; the handle's sole owner becomes `Device`.
        let dev = unsafe { CreateSyntheticPointerDevice(PT_PEN, 1, POINTER_FEEDBACK_DEFAULT) }
            .context("CreateSyntheticPointerDevice(PT_PEN) — needs Windows 10 1809+")?;
        let shared = Arc::new(Mutex::new(PenShared {
            dev: Device(dev),
            state: PenState::default(),
            fail_warned: false,
        }));
        let stop = Arc::new(AtomicBool::new(false));
        // The staleness guard: re-assert the last frame while in range so a stationary pen
        // (native plane: between 100 ms heartbeats; GameStream: indefinitely) never hovers out.
        let refresher = {
            let shared = Arc::clone(&shared);
            let stop = Arc::clone(&stop);
            std::thread::Builder::new()
                .name("pf-pen-refresh".into())
                .spawn(move || {
                    while !stop.load(Ordering::Relaxed) {
                        std::thread::sleep(std::time::Duration::from_millis(REFRESH_MS));
                        let s = &mut *shared.lock().unwrap();
                        if s.state.in_range {
                            inject_pen(s, POINTER_FLAG_UPDATE, false);
                        }
                    }
                })
                .context("spawn pen refresh thread")?
        };
        tracing::info!("virtual pen created (Windows synthetic pointer, PT_PEN)");
        Ok(VirtualPen {
            shared,
            stop,
            refresher: Some(refresher),
            edge_down: false,
            edge_up: false,
            is_new: false,
            frame_dirty: false,
            frame_has_motion: false,
        })
    }
    /// Apply one batch of tracker transitions — same grouping contract as the Linux backend:
    /// `[ProxIn, Motion, TipDown]` is ONE frame (entry lands at its position, in contact),
    /// consecutive `Motion`s split, tip edges own their frames, range-leave is a final
    /// no-`INRANGE` frame.
    pub fn apply_batch(&mut self, transitions: &[PenTransition]) {
        for t in transitions {
            match t {
                PenTransition::ProximityIn { tool } => {
                    self.flush();
                    let mut s = self.shared.lock().unwrap();
                    s.state.in_range = true;
                    s.state.eraser = *tool == PenTool::Eraser;
                    drop(s);
                    self.is_new = true;
                    self.frame_dirty = true;
                }
                PenTransition::Motion { sample } => {
                    if self.frame_has_motion {
                        self.flush();
                    }
                    self.set_axes(sample);
                    self.frame_dirty = true;
                    self.frame_has_motion = true;
                }
                PenTransition::TipDown => {
                    self.shared.lock().unwrap().state.touching = true;
                    self.edge_down = true;
                    self.frame_dirty = true;
                }
                PenTransition::ButtonsChanged { pressed, released } => {
                    let mut s = self.shared.lock().unwrap();
                    if pressed & PEN_BARREL1 != 0 {
                        s.state.barrel = true;
                    }
                    if released & PEN_BARREL1 != 0 {
                        s.state.barrel = false;
                    }
                    drop(s);
                    self.frame_dirty = true;
                }
                PenTransition::TipUp => {
                    self.shared.lock().unwrap().state.touching = false;
                    self.edge_up = true;
                    self.frame_dirty = true;
                    self.flush(); // UP owns its frame (still INRANGE)
                }
                PenTransition::ProximityOut => {
                    self.shared.lock().unwrap().state.in_range = false;
                    self.frame_dirty = true;
                    self.flush(); // final frame without INRANGE
                }
            }
        }
        self.flush();
    }
    fn set_axes(&mut self, s: &PenSample) {
        let mut sh = self.shared.lock().unwrap();
        sh.state.x = s.x;
        sh.state.y = s.y;
        sh.state.pressure = s.pressure;
        sh.state.tilt_deg = s.tilt_deg;
        sh.state.azimuth_deg = s.azimuth_deg;
        sh.state.roll_deg = s.roll_deg;
    }
    fn flush(&mut self) {
        if !self.frame_dirty {
            return;
        }
        let edge = if self.edge_down {
            POINTER_FLAG_DOWN
        } else if self.edge_up {
            POINTER_FLAG_UP
        } else {
            POINTER_FLAG_UPDATE
        };
        inject_pen(&mut self.shared.lock().unwrap(), edge, self.is_new);
        self.edge_down = false;
        self.edge_up = false;
        self.is_new = false;
        self.frame_dirty = false;
        self.frame_has_motion = false;
    }
 }
 impl Drop for VirtualPen {
    fn drop(&mut self) {
        self.stop.store(true, Ordering::Relaxed);
        if let Some(h) = self.refresher.take() {
            let _ = h.join();
        }
        // The device itself dies with `shared` (Device::drop) — Windows releases any held
        // in-range/contact state when the synthetic device is destroyed.
    }
 }
 /// Build + inject one pen frame from tracked state. `edge` is DOWN/UP/UPDATE;
 /// `INRANGE`/`INCONTACT` derive from the state itself.
 fn inject_pen(sh: &mut PenShared, edge: POINTER_FLAGS, is_new: bool) {
    let st = &sh.state;
    let mut flags = edge;
    if st.in_range {
        flags |= POINTER_FLAG_INRANGE;
    }
    if st.touching {
        flags |= POINTER_FLAG_INCONTACT | POINTER_FLAG_FIRSTBUTTON;
    }
    if is_new {
        flags |= POINTER_FLAG_NEW;
    }
    let mut pen_flags = PEN_FLAG_NONE;
    if st.barrel {
        pen_flags |= PEN_FLAG_BARREL;
    }
    if st.eraser {
        pen_flags |= PEN_FLAG_INVERTED;
        if st.touching {
            pen_flags |= PEN_FLAG_ERASER;
        }
    }
    let mut pen_mask = PEN_MASK_PRESSURE;
    // Contact needs a nonzero pressure to ink; hover reports 0 (Windows convention).
    let pressure = if st.touching {
        ((st.pressure as u32 * WIN_PEN_PRESSURE_MAX) / u16::MAX as u32).max(1)
    } else {
        0
    };
    let (mut tilt_x, mut tilt_y) = (0i32, 0i32);
    if st.tilt_deg != PEN_TILT_UNKNOWN && st.azimuth_deg != PEN_ANGLE_UNKNOWN {
        let az = (st.azimuth_deg as f32).to_radians();
        let tilt = st.tilt_deg as f32;
        tilt_x = (tilt * az.sin()).round() as i32;
        tilt_y = (-tilt * az.cos()).round() as i32;
        pen_mask |= PEN_MASK_TILT_X | PEN_MASK_TILT_Y;
    }
    let mut rotation = 0u32;
    if st.roll_deg != PEN_ANGLE_UNKNOWN {
        rotation = (st.roll_deg % 360) as u32;
        pen_mask |= PEN_MASK_ROTATION;
    }
    let pt = to_screen(st.x, st.y);
    let info = POINTER_TYPE_INFO {
        r#type: PT_PEN,
        Anonymous: POINTER_TYPE_INFO_0 {
            penInfo: POINTER_PEN_INFO {
                pointerInfo: POINTER_INFO {
                    pointerType: PT_PEN,
                    pointerId: 0,
                    pointerFlags: flags,
                    ptPixelLocation: pt,
                    ptPixelLocationRaw: pt,
                    ..Default::default()
                },
                penFlags: pen_flags,
                penMask: pen_mask,
                pressure,
                rotation,
                tiltX: tilt_x,
                tiltY: tilt_y,
            },
        },
    };
    // SAFETY: `sh.dev.0` is the live device this wrapper owns; the one-element array is a live
    // stack value the call only reads. Best-effort like every injector write — a transient
    // failure (desktop switch) is healed by the next refresh tick re-asserting state.
    if let Err(e) = unsafe { InjectSyntheticPointerInput(sh.dev.0, &[info]) } {
        if !sh.fail_warned {
            sh.fail_warned = true;
            tracing::warn!(error = %e, "pen inject failed");
        } else {
            tracing::trace!(error = %e, "pen inject failed (transient)");
        }
    } else {
        sh.fail_warned = false;
    }
 }
 /// One live wire-touch contact. `slot` is the SMALL, DENSE pointer id handed to Windows —
 /// synthetic-pointer injection rejects arbitrary large ids, and clients (Moonlight's
 /// `pointerId` especially) send exactly those, so wire ids compact into the lowest free slot
 /// for the contact's lifetime (Apollo's slot-contiguity rule; the on-glass symptom of passing
 /// wire ids through was pen working while every touch silently failed to inject).
 #[derive(Clone, Copy)]
 struct Contact {
    id: u32,
    slot: u32,
    x: f32,
    y: f32,
 }
 /// Lowest slot not held by a live contact.
 fn free_slot(contacts: &[Contact]) -> u32 {
    let mut slot = 0u32;
    while contacts.iter().any(|c| c.slot == slot) {
        slot += 1;
    }
    slot
 }
 /// Windows can inject at most this many simultaneous synthetic touch contacts.
 const MAX_CONTACTS: usize = 10;
 struct TouchShared {
    dev: Device,
    contacts: Vec<Contact>,
    /// First injection failure logs at WARN (an on-glass "touch does nothing" must be
    /// visible in the host log); repeats stay at trace.
    fail_warned: bool,
 }
 /// The `PT_TOUCH` device servicing wire `TouchDown/Move/Up` events — closes the SendInput
 /// touch no-op. Contacts are keyed by the wire's finger id; every frame re-injects the FULL
 /// active set (the synthetic-pointer contract), and a refresh thread re-asserts held contacts
 /// against the ~100 ms staleness auto-lift.
 pub struct SyntheticTouch {
    shared: Arc<Mutex<TouchShared>>,
    stop: Arc<AtomicBool>,
    refresher: Option<std::thread::JoinHandle<()>>,
 }
 impl SyntheticTouch {
    pub fn create() -> Result<SyntheticTouch> {
        // SAFETY: FFI create with by-value args; the handle's sole owner becomes `Device`.
        let dev = unsafe {
            CreateSyntheticPointerDevice(PT_TOUCH, MAX_CONTACTS as u32, POINTER_FEEDBACK_DEFAULT)
        }
        .context("CreateSyntheticPointerDevice(PT_TOUCH) — needs Windows 10 1809+")?;
        let shared = Arc::new(Mutex::new(TouchShared {
            dev: Device(dev),
            contacts: Vec::new(),
            fail_warned: false,
        }));
        let stop = Arc::new(AtomicBool::new(false));
        let refresher = {
            let shared = Arc::clone(&shared);
            let stop = Arc::clone(&stop);
            std::thread::Builder::new()
                .name("pf-touch-refresh".into())
                .spawn(move || {
                    while !stop.load(Ordering::Relaxed) {
                        std::thread::sleep(std::time::Duration::from_millis(REFRESH_MS));
                        let s = &mut *shared.lock().unwrap();
                        if !s.contacts.is_empty() {
                            inject_touch_frame(s, None);
                        }
                    }
                })
                .context("spawn touch refresh thread")?
        };
        tracing::info!("virtual touchscreen created (Windows synthetic pointer, PT_TOUCH)");
        Ok(SyntheticTouch {
            shared,
            stop,
            refresher: Some(refresher),
        })
    }
    /// Apply one wire touch event (`code` = finger id, pixel x/y against the
    /// `flags = (w << 16) | h` reference extent, exactly like `MouseMoveAbs`).
    pub fn apply(&mut self, ev: &InputEvent) {
        let (w, h) = ((ev.flags >> 16) as f32, (ev.flags & 0xFFFF) as f32);
        if (w < 1.0 || h < 1.0) && ev.kind != InputKind::TouchUp {
            return; // the documented zero-extent drop, as for MouseMoveAbs
        }
        let (x, y) = (ev.x as f32 / w.max(1.0), ev.y as f32 / h.max(1.0));
        let s = &mut *self.shared.lock().unwrap();
        match ev.kind {
            InputKind::TouchDown => {
                match s.contacts.iter().position(|c| c.id == ev.code) {
                    Some(i) => (s.contacts[i].x, s.contacts[i].y) = (x, y),
                    None if s.contacts.len() < MAX_CONTACTS => {
                        let slot = free_slot(&s.contacts);
                        s.contacts.push(Contact {
                            id: ev.code,
                            slot,
                            x,
                            y,
                        })
                    }
                    None => return, // beyond the platform max — drop, never evict a live finger
                }
                inject_touch_frame(s, Some((ev.code, POINTER_FLAG_DOWN)));
            }
            InputKind::TouchMove => {
                match s.contacts.iter().position(|c| c.id == ev.code) {
                    Some(i) => {
                        (s.contacts[i].x, s.contacts[i].y) = (x, y);
                        inject_touch_frame(s, None);
                    }
                    // A move for an unknown id (its DOWN was dropped/lost): synthesize the
                    // contact so the stroke self-heals, like the pen plane does.
                    None if s.contacts.len() < MAX_CONTACTS => {
                        let slot = free_slot(&s.contacts);
                        s.contacts.push(Contact {
                            id: ev.code,
                            slot,
                            x,
                            y,
                        });
                        inject_touch_frame(s, Some((ev.code, POINTER_FLAG_DOWN)));
                    }
                    None => {}
                }
            }
            InputKind::TouchUp => {
                let Some(idx) = s.contacts.iter().position(|c| c.id == ev.code) else {
                    return;
                };
                // The UP frame still carries the lifting contact (with UP flags), then it
                // leaves the active set.
                inject_touch_frame(s, Some((ev.code, POINTER_FLAG_UP)));
                s.contacts.remove(idx);
            }
            _ => {}
        }
    }
 }
 impl Drop for SyntheticTouch {
    fn drop(&mut self) {
        self.stop.store(true, Ordering::Relaxed);
        if let Some(h) = self.refresher.take() {
            let _ = h.join();
        }
    }
 }
 /// Inject the full active-contact frame; `edge` marks one contact's DOWN/UP transition (by
 /// WIRE id — everyone else is a held UPDATE). Windows sees the compacted `slot` ids only.
 fn inject_touch_frame(sh: &mut TouchShared, edge: Option<(u32, POINTER_FLAGS)>) {
    let contacts = &sh.contacts;
    if contacts.is_empty() {
        return;
    }
    let mut frame: Vec<POINTER_TYPE_INFO> = Vec::with_capacity(contacts.len());
    for c in contacts {
        let mut flags = POINTER_FLAG_INRANGE | POINTER_FLAG_INCONTACT | POINTER_FLAG_UPDATE;
        if let Some((id, e)) = edge {
            if id == c.id {
                if e == POINTER_FLAG_DOWN {
                    flags = POINTER_FLAG_INRANGE | POINTER_FLAG_INCONTACT | POINTER_FLAG_DOWN;
                } else if e == POINTER_FLAG_UP {
                    flags = POINTER_FLAG_UP; // contact + range end together
                }
            }
        }
        let pt = to_screen(c.x, c.y);
        frame.push(POINTER_TYPE_INFO {
            r#type: PT_TOUCH,
            Anonymous: POINTER_TYPE_INFO_0 {
                touchInfo: POINTER_TOUCH_INFO {
                    pointerInfo: POINTER_INFO {
                        pointerType: PT_TOUCH,
                        pointerId: c.slot,
                        pointerFlags: flags,
                        ptPixelLocation: pt,
                        ptPixelLocationRaw: pt,
                        ..Default::default()
                    },
                    touchFlags: TOUCH_FLAG_NONE,
                    touchMask: TOUCH_MASK_NONE,
                    ..Default::default()
                },
            },
        });
    }
    // SAFETY: `sh.dev.0` is the live owned device; `frame` is a live Vec the call only reads.
    // Best-effort — a transient failure heals on the next event/refresh re-assertion.
    if let Err(e) = unsafe { InjectSyntheticPointerInput(sh.dev.0, &frame) } {
        if !sh.fail_warned {
            sh.fail_warned = true;
            tracing::warn!(error = %e, contacts = frame.len(), "touch inject failed");
        } else {
            tracing::trace!(error = %e, "touch inject failed (transient)");
        }
    } else {
        sh.fail_warned = false;
    }
 }
@@ -27,10 +27,10 @@ use windows::Win32::System::StationsAndDesktops::{
 use windows::Win32::UI::Input::KeyboardAndMouse::{
    GetKeyboardLayout, MapVirtualKeyExW, SendInput, HKL, INPUT, INPUT_0, INPUT_KEYBOARD,
    INPUT_MOUSE, KEYBDINPUT, KEYEVENTF_EXTENDEDKEY, KEYEVENTF_KEYUP, KEYEVENTF_SCANCODE,
-    MAPVK_VK_TO_VSC_EX, MOUSEEVENTF_ABSOLUTE, MOUSEEVENTF_HWHEEL, MOUSEEVENTF_LEFTDOWN,
+    KEYEVENTF_UNICODE, MAPVK_VK_TO_VSC_EX, MOUSEEVENTF_ABSOLUTE, MOUSEEVENTF_HWHEEL,
-    MOUSEEVENTF_LEFTUP, MOUSEEVENTF_MIDDLEDOWN, MOUSEEVENTF_MIDDLEUP, MOUSEEVENTF_MOVE,
+    MOUSEEVENTF_LEFTDOWN, MOUSEEVENTF_LEFTUP, MOUSEEVENTF_MIDDLEDOWN, MOUSEEVENTF_MIDDLEUP,
-    MOUSEEVENTF_RIGHTDOWN, MOUSEEVENTF_RIGHTUP, MOUSEEVENTF_VIRTUALDESK, MOUSEEVENTF_WHEEL,
+    MOUSEEVENTF_MOVE, MOUSEEVENTF_RIGHTDOWN, MOUSEEVENTF_RIGHTUP, MOUSEEVENTF_VIRTUALDESK,
-    MOUSEEVENTF_XDOWN, MOUSEEVENTF_XUP, MOUSEINPUT, VIRTUAL_KEY,
+    MOUSEEVENTF_WHEEL, MOUSEEVENTF_XDOWN, MOUSEEVENTF_XUP, MOUSEINPUT, VIRTUAL_KEY,
 };
 use windows::Win32::UI::WindowsAndMessaging::{
    GetForegroundWindow, GetSystemMetrics, GetWindowThreadProcessId, SM_CXVIRTUALSCREEN,
@@ -46,6 +46,11 @@ const XBUTTON2: u32 = 0x0002;
 pub struct SendInputInjector {
    desktop: Option<HDESK>,
    /// PT_TOUCH synthetic device, created on the first wire-touch event (a session that never
    /// touches never creates one). `None` after a failed create (pre-1809) — touch then stays
    /// the historical no-op.
    touch: Option<crate::pen::SyntheticTouch>,
    touch_failed: bool,
 }
 // SAFETY: `SendInputInjector` holds only an `Option<HDESK>` (a desktop handle). The host creates
@@ -58,7 +63,11 @@ unsafe impl Send for SendInputInjector {}
 impl SendInputInjector {
    pub fn open() -> Result<Self> {
-        let mut me = Self { desktop: None };
+        let mut me = Self {
            desktop: None,
            touch: None,
            touch_failed: false,
        };
        me.reattach_input_desktop(); // best-effort
        tracing::info!("SendInput injector ready (Win32 KeyboardAndMouse)");
        Ok(me)
@@ -297,15 +306,60 @@ impl InputInjector for SendInputInjector {
                };
                self.send(&[key(ki)])
            }
-            // Gamepad goes through the XUSB backend. Touch: no SendInput equivalent -> no-op.
+            InputKind::TextInput => {
                // Committed IME text: one Unicode scalar per event, injected as
                // `KEYEVENTF_UNICODE` packets (wScan = UTF-16 unit, no scancode/layout involved
                // — the receiving app gets the character verbatim via WM_CHAR). An astral-plane
                // scalar (emoji) is its surrogate pair, each unit down+up in order — exactly how
                // Windows expects supplementary characters from unicode injection.
                let Some(ch) = char::from_u32(event.code) else {
                    return Ok(()); // lone surrogate / out of range — drop
                };
                if ch.is_control() {
                    return Ok(()); // control chars ride the VK path (Enter/Backspace/Tab)
                }
                let mut units = [0u16; 2];
                let mut inputs: Vec<INPUT> = Vec::with_capacity(4);
                for &unit in ch.encode_utf16(&mut units).iter() {
                    for flags in [KEYEVENTF_UNICODE, KEYEVENTF_UNICODE | KEYEVENTF_KEYUP] {
                        inputs.push(key(KEYBDINPUT {
                            wVk: VIRTUAL_KEY(0),
                            wScan: unit,
                            dwFlags: flags,
                            time: 0,
                            dwExtraInfo: 0,
                        }));
                    }
                }
                self.send(&inputs)
            }
            // Gamepad goes through the XUSB backend.
            InputKind::GamepadButton
            | InputKind::GamepadAxis
            | InputKind::GamepadState
            | InputKind::GamepadRemove
-            | InputKind::GamepadArrival
+            | InputKind::GamepadArrival => Ok(()),
-            | InputKind::TouchDown
+            // Wire touch → the PT_TOUCH synthetic pointer device (design/pen-tablet-input.md
-            | InputKind::TouchMove
+            // §6; closes the historical SendInput no-op). Lazily created — a session that never
-            | InputKind::TouchUp => Ok(()),
+            // touches never creates one; a pre-1809 create failure latches back to the no-op.
            InputKind::TouchDown | InputKind::TouchMove | InputKind::TouchUp => {
                if self.touch.is_none() && !self.touch_failed {
                    match crate::pen::SyntheticTouch::create() {
                        Ok(t) => self.touch = Some(t),
                        Err(e) => {
                            self.touch_failed = true;
                            tracing::warn!(
                                error = %format!("{e:#}"),
                                "touch: synthetic pointer unavailable — wire touch stays a no-op"
                            );
                        }
                    }
                }
                if let Some(t) = self.touch.as_mut() {
                    t.apply(event);
                }
                Ok(())
            }
        }
    }
 }
@@ -174,6 +174,73 @@ pub fn default_backend() -> Backend {
    Backend::Unsupported
 }
 /// Whether the session's inject backend can type **committed text**
 /// ([`InputKind::TextInput`] — see `HOST_CAP_TEXT_INPUT`): Windows always (`KEYEVENTF_UNICODE`);
 /// Linux only on the wlroots backend (a dedicated virtual keyboard with a dynamically-grown
 /// Unicode keymap) — KWin fake-input/libei/gamescope can only press keycodes of the host layout.
 /// Consulted at Welcome time to advertise the cap; a mid-session backend switch away from a
 /// capable one just degrades to dropped text events (input is lossy by design).
 #[cfg(target_os = "windows")]
 pub fn text_input_supported() -> bool {
    true
 }
 /// See the Windows variant: Linux types text only through the wlroots virtual-keyboard backend.
 #[cfg(target_os = "linux")]
 pub fn text_input_supported() -> bool {
    matches!(default_backend(), Backend::WlrVirtual)
 }
 /// No injector ⇒ no text.
 #[cfg(not(any(target_os = "linux", target_os = "windows")))]
 pub fn text_input_supported() -> bool {
    false
 }
 /// Whether this host can inject full-fidelity stylus input (`HOST_CAP_PEN` —
 /// design/pen-tablet-input.md): Linux only, via the [`pen::VirtualPen`] uinput tablet, so the
 /// probe is "can we open /dev/uinput" (the same permission the virtual gamepads need) plus the
 /// `PUNKTFUNK_PEN=0` operator kill-switch. Consulted at Welcome time; clients without the bit
 /// keep folding pen into touch/pointer. Windows PT_PEN synthetic pointers are the design's P3.
 #[cfg(target_os = "linux")]
 pub fn pen_supported() -> bool {
    if std::env::var("PUNKTFUNK_PEN").as_deref() == Ok("0") {
        return false;
    }
    // SAFETY: 'static NUL-terminated path literal; `open` returns a fresh fd (or -1) and
    // retains nothing.
    let fd = unsafe {
        libc::open(
            c"/dev/uinput".as_ptr(),
            libc::O_RDWR | libc::O_NONBLOCK | libc::O_CLOEXEC,
        )
    };
    if fd < 0 {
        return false;
    }
    // SAFETY: `fd >= 0` is the fd opened above, owned by no one else; closed exactly once here.
    unsafe { libc::close(fd) };
    true
 }
 /// Windows: pen (and touch) inject via synthetic pointer devices — available on Win10 1809+,
 /// probed by actually creating (and immediately destroying) a PT_PEN device. Same
 /// `PUNKTFUNK_PEN=0` kill-switch as Linux. The probe result also stands in for PT_TOUCH
 /// (both APIs arrived together in 1809).
 #[cfg(target_os = "windows")]
 pub fn pen_supported() -> bool {
    if std::env::var("PUNKTFUNK_PEN").as_deref() == Ok("0") {
        return false;
    }
    pen::synthetic_pen_available()
 }
 /// See the Linux/Windows variants — no pen injection elsewhere.
 #[cfg(not(any(target_os = "linux", target_os = "windows")))]
 pub fn pen_supported() -> bool {
    false
 }
 #[path = "inject/service.rs"]
 mod service;
 pub use service::InjectorService;
@@ -339,6 +406,30 @@ pub mod gamepad {
        pub fn pump_rumble(&mut self, _send: impl FnMut(u16, u16, u16)) {}
    }
 }
 /// Linux: the "Punktfunk Pen" uinput virtual tablet (design/pen-tablet-input.md §5) — the
 /// per-session stylus device the native pen plane injects through.
 #[cfg(target_os = "linux")]
 #[path = "inject/linux/pen.rs"]
 pub mod pen;
 /// Windows: PT_PEN/PT_TOUCH synthetic pointer devices (design/pen-tablet-input.md §6).
 /// `pen::VirtualPen` here is the PT_PEN device; `pen::SyntheticTouch` backs the SendInput
 /// injector's wire-touch path.
 #[cfg(target_os = "windows")]
 #[path = "inject/windows/pointer_windows.rs"]
 pub mod pen;
 /// Stub — pen injection needs the Linux uinput tablet or Windows synthetic pointers;
 /// `pen_supported()` is false here, so no host advertises the cap and no batches arrive.
 #[cfg(not(any(target_os = "linux", target_os = "windows")))]
 pub mod pen {
    use anyhow::{bail, Result};
    pub struct VirtualPen;
    impl VirtualPen {
        pub fn create() -> Result<VirtualPen> {
            bail!("no pen injection backend on this platform")
        }
        pub fn apply_batch(&mut self, _transitions: &[punktfunk_core::quic::PenTransition]) {}
    }
 }
 #[cfg(target_os = "linux")]
 #[path = "inject/linux/kwin_fake_input.rs"]
 mod kwin_fake_input;
@@ -0,0 +1,232 @@
 //! Client-side cursor rendering (design/remote-desktop-sweep.md M2): the host forwards the
 //! pointer's SHAPE (reliable control stream, cached by serial) and per-frame STATE (lossy
 //! `0xD0` — position/visibility), and WE draw it as a real OS cursor — pointer feel stops
 //! paying the video round-trip (the Parsec/RDP model). Active only when the session
 //! negotiated it (`HOST_CAP_CURSOR` in the Welcome — the host stopped compositing then) and
 //! only applied while the DESKTOP mouse model is engaged: under capture the pointer is
 //! relative-locked (SDL hides it) and games draw their own cursor in-frame.
 //!
 //! The host sends the bitmap in host-FRAMEBUFFER pixels, whose size tracks the host virtual
 //! display's DPI scaling (32 px at 100%, 96 px at 300%). Drawn 1:1 it balloons on a high-DPI
 //! host; instead we scale it by the SAME aspect-fit factor the video is drawn at
 //! (`min(window_px/mode)`), so the pointer stays sized to the streamed desktop at any host
 //! scaling. SDL cursors are fixed-size from their surface (no draw-time scaling), so we cache
 //! shapes RAW and resample per install — rebuilding when the serial OR the fit changes.
 use punktfunk_core::client::NativeClient;
 use punktfunk_core::quic::{CursorState, HOST_CAP_CURSOR};
 use sdl3::mouse::{Cursor, MouseUtil, SystemCursor};
 use sdl3::pixels::PixelFormat;
 use sdl3::surface::Surface;
 use std::collections::HashMap;
 use std::time::Duration;
 /// Shape serials cached at most — cursors cycle through a handful of shapes (arrow, I-beam,
 /// resize…); a runaway host can't grow the map past this (the cache resets, shapes re-arrive
 /// on the reliable stream via the serial-miss path).
 const SHAPE_CACHE_MAX: usize = 64;
 /// A forwarded cursor shape held RAW (host-framebuffer-pixel bytes + hotspot), so it can be
 /// rebuilt into a scaled OS cursor whenever the video-fit changes (a window resize). Caching a
 /// fixed-size `Cursor` instead would freeze the pointer at its build-time size.
 struct RawShape {
    rgba: Vec<u8>,
    w: u32,
    h: u32,
    hot_x: u32,
    hot_y: u32,
 }
 pub struct CursorChannel {
    /// The Welcome carried `HOST_CAP_CURSOR` — the host forwards instead of compositing.
    negotiated: bool,
    /// Serial → raw forwarded shape (bounded by [`SHAPE_CACHE_MAX`]).
    shapes: HashMap<u32, RawShape>,
    /// The serial + fit scale the currently-installed OS cursor was built at (`None` =
    /// default/system cursor). A change in EITHER forces a rebuild.
    installed: Option<(u32, f32)>,
    /// Keeps the installed `Cursor` alive — SDL requires it to outlive its `set()`.
    installed_cursor: Option<Cursor>,
    /// Latest `0xD0` state (latest-wins across a drained batch).
    state: Option<CursorState>,
 }
 impl CursorChannel {
    pub fn new(connector: &NativeClient) -> CursorChannel {
        let negotiated = connector.host_caps() & HOST_CAP_CURSOR != 0;
        if negotiated {
            tracing::info!("cursor channel negotiated — host cursor renders locally");
        }
        CursorChannel {
            negotiated,
            shapes: HashMap::new(),
            installed: None,
            installed_cursor: None,
            state: None,
        }
    }
    /// Whether the host forwards the cursor this session (it no longer composites one).
    pub fn negotiated(&self) -> bool {
        self.negotiated
    }
    /// The latest drained `0xD0` state — the run loop reads `relative_hint` off it for the
    /// M3 host-driven mode flip (and `x`/`y` as the reappear position when leaving relative).
    pub fn state(&self) -> Option<CursorState> {
        self.state
    }
    /// Drain the two planes and apply the newest state — once per run-loop iteration.
    /// `desktop_active` = the desktop mouse model is engaged (captured + desktop): only then
    /// do we own the local cursor's shape/visibility; under capture SDL's relative mode owns
    /// it, and released the system cursor must look normal. `fit_scale` is host-framebuffer
    /// pixels → window pixels (the aspect-fit factor the video is drawn at); the shape is
    /// resampled by it so the pointer matches the streamed desktop at any host DPI.
    pub fn pump(
        &mut self,
        connector: &NativeClient,
        mouse: &MouseUtil,
        desktop_active: bool,
        fit_scale: f32,
    ) {
        if !self.negotiated {
            return;
        }
        while let Ok(shape) = connector.next_cursor_shape(Duration::ZERO) {
            if self.shapes.len() >= SHAPE_CACHE_MAX {
                // Degenerate host: reset — live shapes re-install via the serial-miss path.
                self.shapes.clear();
                self.installed = None;
            }
            let (w, h) = (shape.w as u32, shape.h as u32);
            if w == 0 || h == 0 || shape.rgba.len() < (w * h * 4) as usize {
                tracing::warn!(w, h, "cursor shape malformed — ignored");
                continue;
            }
            // A re-sent serial replaces its entry; force re-install if it's current.
            if matches!(self.installed, Some((s, _)) if s == shape.serial) {
                self.installed = None;
            }
            self.shapes.insert(
                shape.serial,
                RawShape {
                    rgba: shape.rgba,
                    w,
                    h,
                    hot_x: shape.hot_x as u32,
                    hot_y: shape.hot_y as u32,
                },
            );
        }
        while let Ok(st) = connector.next_cursor_state(Duration::ZERO) {
            self.state = Some(st); // latest wins
        }
        if !desktop_active {
            // Capture mode / released: hand the cursor back to the system default so a
            // released pointer over the window doesn't wear the host's shape.
            if self.installed.take().is_some() {
                if let Ok(c) = Cursor::from_system(SystemCursor::Arrow) {
                    c.set();
                    self.installed_cursor = Some(c); // keep it alive past set()
                }
            }
            return;
        }
        let Some(st) = self.state else { return };
        if st.visible() && self.installed != Some((st.serial, fit_scale)) {
            if let Some(shape) = self.shapes.get(&st.serial) {
                match build_scaled_cursor(shape, fit_scale) {
                    Ok(cursor) => {
                        cursor.set();
                        self.installed = Some((st.serial, fit_scale));
                        self.installed_cursor = Some(cursor); // outlive set()
                    }
                    Err(e) => tracing::warn!(error = %e, w = shape.w, h = shape.h,
                        "cursor shape rejected by SDL — keeping the previous cursor"),
                }
            }
            // Serial miss: the (reliable) shape hasn't landed yet — keep the previous
            // cursor for the RTT rather than flashing default.
        }
        // Visibility follows the host (a host app hid its pointer ⇒ ours hides too). Queried,
        // not shadowed, so apply_capture's own show/hide calls can never desync us.
        if mouse.is_cursor_showing() != st.visible() {
            mouse.show_cursor(st.visible());
        }
    }
 }
 /// Resample a raw shape by `fit_scale` and build an SDL color cursor from it. The hotspot scales
 /// with the bitmap so the click point stays true. `fit_scale <= 0` (or a degenerate result) is
 /// clamped so we always hand SDL a ≥1×1 surface.
 fn build_scaled_cursor(shape: &RawShape, fit_scale: f32) -> Result<Cursor, String> {
    let scale = if fit_scale.is_finite() && fit_scale > 0.0 {
        fit_scale
    } else {
        1.0
    };
    let dw = ((shape.w as f32 * scale).round() as u32).max(1);
    let dh = ((shape.h as f32 * scale).round() as u32).max(1);
    let hot_x = ((shape.hot_x as f32 * scale).round() as u32).min(dw - 1) as i32;
    let hot_y = ((shape.hot_y as f32 * scale).round() as u32).min(dh - 1) as i32;
    if dw == shape.w && dh == shape.h {
        // 1:1 fit (mode == window) — no resample, feed the bytes straight through.
        let mut data = shape.rgba.clone();
        let surf = Surface::from_data(
            &mut data,
            shape.w,
            shape.h,
            shape.w * 4,
            PixelFormat::RGBA32,
        )
        .map_err(|e| e.to_string())?;
        return Cursor::from_surface(&surf, hot_x, hot_y).map_err(|e| e.to_string());
    }
    let mut scaled = resample_rgba(&shape.rgba, shape.w, shape.h, dw, dh);
    let surf = Surface::from_data(&mut scaled, dw, dh, dw * 4, PixelFormat::RGBA32)
        .map_err(|e| e.to_string())?;
    Cursor::from_surface(&surf, hot_x, hot_y).map_err(|e| e.to_string())
 }
 /// Area-average resample of a straight-alpha RGBA bitmap `(sw×sh) → (dw×dh)`. Averaging is done on
 /// PREMULTIPLIED colour (weighting each source texel by its alpha) so transparent-pixel colour
 /// can't bleed into the fringe, then un-premultiplied back to straight alpha. Handles both down-
 /// and up-scale; for a cursor the common case is a downscale (high-DPI host → smaller pointer).
 fn resample_rgba(src: &[u8], sw: u32, sh: u32, dw: u32, dh: u32) -> Vec<u8> {
    let mut out = vec![0u8; (dw * dh * 4) as usize];
    let fx = sw as f32 / dw as f32;
    let fy = sh as f32 / dh as f32;
    for dy in 0..dh {
        let sy0 = (dy as f32 * fy).floor() as u32;
        let sy1 = (((dy + 1) as f32 * fy).ceil() as u32).clamp(sy0 + 1, sh);
        for dx in 0..dw {
            let sx0 = (dx as f32 * fx).floor() as u32;
            let sx1 = (((dx + 1) as f32 * fx).ceil() as u32).clamp(sx0 + 1, sw);
            let (mut r, mut g, mut b, mut a_sum, mut n) = (0f32, 0f32, 0f32, 0f32, 0f32);
            for sy in sy0..sy1 {
                for sx in sx0..sx1 {
                    let i = ((sy * sw + sx) * 4) as usize;
                    let a = src[i + 3] as f32 / 255.0;
                    r += src[i] as f32 * a;
                    g += src[i + 1] as f32 * a;
                    b += src[i + 2] as f32 * a;
                    a_sum += a;
                    n += 1.0;
                }
            }
            let di = ((dy * dw + dx) * 4) as usize;
            if a_sum > 0.0 {
                out[di] = (r / a_sum).round().clamp(0.0, 255.0) as u8;
                out[di + 1] = (g / a_sum).round().clamp(0.0, 255.0) as u8;
                out[di + 2] = (b / a_sum).round().clamp(0.0, 255.0) as u8;
                out[di + 3] = (a_sum / n * 255.0).round().clamp(0.0, 255.0) as u8;
            }
            // else fully transparent — already zero-filled.
        }
    }
    out
 }
@@ -133,6 +133,20 @@ impl Capture {
        Some(self.desktop)
    }
    /// Set the mouse model directly (the M3 host-driven flip — `relative_hint` says a host
    /// app grabbed/hid the pointer, so run relative; hint clear = back to absolute). Same
    /// gating and motion hygiene as [`toggle_desktop`](Self::toggle_desktop); returns whether
    /// the model actually changed.
    pub fn set_desktop(&mut self, on: bool) -> bool {
        if !self.abs_ok || self.desktop == on {
            return false;
        }
        self.desktop = on;
        self.pending_rel = (0, 0);
        self.pending_abs = None;
        true
    }
    /// Whether a regained focus should re-engage: yes unless the user released
    /// deliberately (the chord keeps its meaning across an Alt-Tab).
    pub fn should_reengage(&self) -> bool {
@@ -16,6 +16,8 @@
 #[cfg(any(target_os = "linux", windows))]
 pub mod csc;
 #[cfg(any(target_os = "linux", windows))]
 pub mod cursor;
 #[cfg(windows)]
 pub mod d3d11;
 #[cfg(target_os = "linux")]
@@ -233,6 +233,19 @@ struct StreamState {
    /// window-normalized position must be re-based onto the content rect). `None` until
    /// the first frame; touches before then have nothing to map onto and are dropped.
    last_video: Option<(u32, u32)>,
    /// Client-side cursor rendering (M2 cursor channel) — created with the connector; inert
    /// when the host didn't negotiate the channel.
    cursor_chan: Option<crate::cursor::CursorChannel>,
    /// Last observed `relative_hint` (M3): the auto-flip fires on CHANGES only, so it never
    /// fights a user who chorded away from the hinted model.
    last_hint: Option<bool>,
    /// The user flipped the model manually (⌃⌥⇧M) — the standing hint stops driving until
    /// the HOST's intent next changes (a fresh hint edge clears this and applies).
    hint_override: bool,
    /// Last `CursorRenderMode.client_draws` told to the host (§8 mid-stream render flip);
    /// `None` = nothing sent yet. Edge-detected each iteration from the live mouse model, so
    /// the chord, the M3 auto-flip, and engage/release all reconcile through one path.
    sent_client_draws: Option<bool>,
 }
 impl StreamState {
@@ -263,6 +276,10 @@ impl StreamState {
            frames: wake_rx,
            connector: None,
            capture: None,
            cursor_chan: None,
            last_hint: None,
            hint_override: false,
            sent_client_draws: None,
            force_software,
            canceled: false,
            ready_announced: false,
@@ -554,12 +571,15 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
                    // Mouse model flip (capture ⇄ desktop) — applies immediately when
                    // engaged; a released stream just changes what the next engage does.
                    if chord && sc == Scancode::M {
-                        if let Some(cap) = stream.as_mut().and_then(|s| s.capture.as_mut()) {
+                        if let Some(st) = stream.as_mut() {
                            let mut flipped = false;
                            if let Some(cap) = st.capture.as_mut() {
                                match cap.toggle_desktop() {
                                    Some(desktop) => {
                                        if cap.captured() {
                                            apply_capture(&mut window, &mouse, true, desktop);
                                        }
                                        flipped = true;
                                        tracing::info!(desktop, "chord: mouse mode");
                                    }
                                    None => tracing::info!(
@@ -568,6 +588,12 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
                                    ),
                                }
                            }
                            // A manual flip outranks the standing hint until the host's
                            // intent next CHANGES (M3 — the hint edge clears this).
                            if flipped {
                                st.hint_override = true;
                            }
                        }
                        continue;
                    }
                    if chord && sc == Scancode::D {
@@ -744,6 +770,75 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
        if let Some(cap) = stream.as_mut().and_then(|s| s.capture.as_mut()) {
            cap.flush_motion();
        }
        // Cursor channel (M2): drain forwarded shape/state and drive the local OS cursor —
        // only meaningful in the desktop mouse model (capture's relative lock hides it).
        if let Some(st) = stream.as_mut() {
            // Host-framebuffer px → window px: the aspect-fit factor the video is drawn at
            // (same `min(surface/content)` as `finger_to_content`). The forwarded pointer is
            // resampled by it so a high-DPI host's oversized bitmap lands sized to the streamed
            // desktop rather than ballooning. 1:1 until the first frame gives `last_video`.
            let fit_scale = st.last_video.map_or(1.0, |(vw, vh)| {
                let (pw, ph) = window.size_in_pixels();
                (pw as f32 / vw.max(1) as f32).min(ph as f32 / vh.max(1) as f32)
            });
            if let (Some(chan), Some(c)) = (st.cursor_chan.as_mut(), st.connector.as_ref()) {
                let desktop_active = st
                    .capture
                    .as_ref()
                    .is_some_and(|cap| cap.captured() && cap.desktop());
                chan.pump(c, &mouse, desktop_active, fit_scale);
                // §8 mid-stream render flip: tell the host who renders the pointer whenever
                // the local model changes. Desktop-active = we draw it (host excludes +
                // forwards); anything else — the capture model OR a released pointer — the
                // host composites it into the video (full fidelity, the pre-channel look).
                // One edge-detected reconciler covers the chord, the M3 auto-flip, and
                // engage/release alike.
                if chan.negotiated() && st.sent_client_draws != Some(desktop_active) {
                    st.sent_client_draws = Some(desktop_active);
                    let _ = c.set_cursor_render(desktop_active);
                }
            }
            // M3 — host-driven mode flip: `relative_hint` set = a host app grabbed/hid the
            // pointer (run captured relative, like a game expects); clear = the desktop is
            // back (return to absolute, local cursor reappearing at the host's position).
            // Edge-triggered so a user's manual chord isn't fought: the override latch
            // holds until the HOST's intent next changes.
            let hint_state = st.cursor_chan.as_ref().and_then(|ch| ch.state());
            if let Some(hs) = hint_state {
                let hint = hs.relative_hint();
                if st.last_hint != Some(hint) {
                    st.last_hint = Some(hint);
                    st.hint_override = false;
                }
                if !st.hint_override {
                    let video = st.last_video;
                    if let Some(cap) = st.capture.as_mut() {
                        // Desired model: hint ⇒ capture (desktop off); clear ⇒ desktop on.
                        if cap.captured() && cap.set_desktop(!hint) {
                            apply_capture(&mut window, &mouse, true, cap.desktop());
                            if cap.desktop() {
                                // Reappear where the host last had the pointer, so the
                                // hand-back is seamless (Parsec's positionX/Y idea).
                                if let Some(video) = video {
                                    let (wx, wy) = content_to_window(
                                        window.size(),
                                        window.size_in_pixels(),
                                        video,
                                        hs.x,
                                        hs.y,
                                    );
                                    mouse.warp_mouse_in_window(&window, wx, wy);
                                }
                            }
                            tracing::info!(
                                desktop = cap.desktop(),
                                "host cursor hint: mouse model flipped"
                            );
                        }
                    }
                }
            }
        }
        // Text input follows the overlay's editing state (edge-triggered).
        let want_text = overlay.as_ref().is_some_and(|o| o.text_input_active());
@@ -888,6 +983,7 @@ fn run_inner(mut opts: SessionOpts, mut mode: ModeCtl) -> Result<Option<Outcome>
                    cap.engage(); // capture engages when the stream starts (ui_stream parity)
                    apply_capture(&mut window, &mouse, true, cap.desktop());
                    st.capture = Some(cap);
                    st.cursor_chan = Some(crate::cursor::CursorChannel::new(&c));
                    st.connector = Some(c);
                    if let Some(f) = opts.on_connected.as_mut() {
                        f(fingerprint);
@@ -1666,6 +1762,32 @@ fn finger_to_content(
    (cx.round() as i32, cy.round() as i32, dw as u32, dh as u32)
 }
 /// The inverse direction of [`finger_to_content`] for the M3 reappear warp: a HOST-frame pixel
 /// (`video` space — what `CursorState` carries) → LOGICAL window coordinates (what
 /// `warp_mouse_in_window` takes). Maps through the aspect-fit letterbox (physical), then
 /// physical → logical via the window's pixel density; out-of-range host coords clamp into the
 /// content rect so the warp always lands on the video.
 fn content_to_window(
    logical: (u32, u32),
    surface: (u32, u32),
    video: (u32, u32),
    x: i32,
    y: i32,
 ) -> (f32, f32) {
    let (pw, ph) = (f64::from(surface.0), f64::from(surface.1));
    let (vw, vh) = (f64::from(video.0.max(1)), f64::from(video.1.max(1)));
    let scale = (pw / vw).min(ph / vh);
    let (dw, dh) = ((vw * scale).max(1.0), (vh * scale).max(1.0));
    let ox = (pw - dw) / 2.0;
    let oy = (ph - dh) / 2.0;
    let px = ox + (f64::from(x).clamp(0.0, vw - 1.0)) * scale;
    let py = oy + (f64::from(y).clamp(0.0, vh - 1.0)) * scale;
    // Physical → logical (HiDPI): the ratio of the window's logical size to its pixel size.
    let lx = px * f64::from(logical.0) / pw.max(1.0);
    let ly = py * f64::from(logical.1) / ph.max(1.0);
    (lx as f32, ly as f32)
 }
 /// The presenter's share of the unified stats window — folded into each printed line.
 #[derive(Default)]
 struct PresentedWindow {
@@ -1763,6 +1885,30 @@ fn stats_text(
 mod tests {
    use super::*;
    #[test]
    fn content_to_window_inverts_the_letterbox() {
        // 1920×1080 video letterboxed in a 1600×1200 (4:3) window at 2× HiDPI: pillarless
        // top/bottom bars — scale = 1600/1920, dh = 900, oy = 150 (physical).
        let logical = (800u32, 600u32);
        let surface = (1600u32, 1200u32);
        let video = (1920u32, 1080u32);
        // The host-frame center must land at the logical window center.
        let (wx, wy) = content_to_window(logical, surface, video, 960, 540);
        assert!((wx - 400.0).abs() < 1.0, "wx = {wx}");
        assert!((wy - 300.0).abs() < 1.0, "wy = {wy}");
        // Roundtrip through the forward mapping: normalized window pos → the same host
        // content-rect pixel (finger_to_content returns content-RECT coords, i.e. the
        // host pixel scaled by the letterbox factor).
        let (nx, ny) = (wx / logical.0 as f32, wy / logical.1 as f32);
        let (cx, cy, dw, dh) = finger_to_content(surface, video, nx, ny);
        assert_eq!((dw, dh), (1600, 900));
        assert!((cx - 800).abs() <= 1, "cx = {cx}"); // 960 * (1600/1920)
        assert!((cy - 450).abs() <= 1, "cy = {cy}"); // 540 * ( 900/1080)
                                                     // Out-of-range host coords clamp into the video, never the bars.
        let (_, wy_clamped) = content_to_window(logical, surface, video, 0, 10_000);
        assert!(wy_clamped <= 300.0 + 225.0 + 1.0, "wy = {wy_clamped}"); // ≤ bottom of content
    }
    #[test]
    fn resize_decision_follows_the_d2_discipline() {
        let t0 = Instant::now();
@@ -72,13 +72,14 @@ pub use session::{session_epoch, try_recover_session};
 #[path = "vdisplay/routing.rs"]
 pub(crate) mod routing;
 pub use routing::{
-    apply_input_env, restore_managed_session, restore_takeover_on_startup, start_restore_worker,
+    apply_input_env, managed_session_available, restore_managed_session,
-    wants_dedicated_game_session,
+    restore_takeover_on_startup, start_restore_worker, wants_dedicated_game_session,
 };
 #[cfg(target_os = "linux")]
 pub use routing::{
-    cancel_pending_tv_restore, dedicated_game_exited, launch_into_gamescope_session,
+    cancel_pending_tv_restore, dedicated_game_exited, gamescope_xwayland_cursor_targets,
-    launch_is_nested, steam_appid_from_launch, watch_steam_game_exit,
+    launch_into_gamescope_session, launch_is_nested, steam_appid_from_launch,
    watch_steam_game_exit,
 };
 /// Compositors punktfunk knows how to drive (plan §6).
@@ -70,6 +70,12 @@ pub struct VirtualOutput {
    /// Linux-only (the keep-alive pool is Linux).
    #[cfg(target_os = "linux")]
    pub pool_gen: Option<u64>,
    /// The backend created the output at a SACRIFICIAL mode and the producer will renegotiate the
    /// live stream to `preferred_mode`'s dims (KWin's screencast only rebuilds its format offer —
    /// and thus its refresh cap — on a size change while recording; see kwin.rs `create`). The
    /// capturer must hold frames until that renegotiation lands. Linux-only.
    #[cfg(target_os = "linux")]
    pub expect_exact_dims: bool,
 }
 impl VirtualOutput {
@@ -93,6 +99,8 @@ impl VirtualOutput {
            reused_gen: None,
            #[cfg(target_os = "linux")]
            pool_gen: None,
            #[cfg(target_os = "linux")]
            expect_exact_dims: false,
        }
    }
 }
@@ -134,6 +142,25 @@ pub trait VirtualDisplay: Send {
    /// the backend's default EDID. Default: no-op — only the Windows pf-vdisplay backend can mint
    /// per-monitor EDIDs today (the Linux compositors' virtual outputs take no EDID from us).
    fn set_client_hdr(&mut self, _hdr: Option<punktfunk_core::quic::HdrMeta>) {}
    /// Ask the backend for an OUT-OF-BAND cursor on the created output (the cursor channel):
    /// the compositor/OS stops compositing the pointer into captured frames and the capture
    /// layer surfaces shape/position separately. Carried on the backend instance; set once
    /// before [`create`](Self::create) (both session paths pass `cursor_forward`). Off = the
    /// compositor EMBEDS the pointer into frames — zero host-side cursor work, the pre-channel
    /// path — which is what every session without the negotiated cursor cap gets (Moonlight /
    /// GameStream / legacy clients / capture-mode starts), mirroring the Windows no-regression
    /// gate. Implementations: Windows pf-vdisplay (IddCx hardware cursor, driver proto v5);
    /// KWin (zkde `pointer` metadata vs embedded); Mutter (`cursor-mode` metadata vs embedded);
    /// wlroots/hyprland (portal `CursorMode`). Default: no-op (gamescope has no cursor either
    /// way — see the Phase C source).
    fn set_hw_cursor(&mut self, _on: bool) {}
    /// The out-of-band-cursor request currently set (see [`set_hw_cursor`](Self::set_hw_cursor)).
    /// The registry includes it in the keep-alive REUSE key: a kept embedded-pointer display can
    /// never serve a cursor-channel session (its stream has no cursor metadata to forward) nor
    /// vice versa (the pointer would be missing from frames).
    fn hw_cursor(&self) -> bool {
        false
    }
    /// The stable identity slot the backend resolved for the most recent [`create`](Self::create) —
    /// the per-client id the identity policy assigned (`Some`), or `None` for shared/anonymous. The
    /// registry reads it right after `create` to key the display's group **arrangement** (manual
@@ -253,6 +253,7 @@ impl VirtualDisplay for GamescopeDisplay {
                ownership: DisplayOwnership::External,
                reused_gen: None,
                pool_gen: None,
                expect_exact_dims: false,
            });
        }
        check_gamescope_version(); // diagnostic only — warns on known-deadlock-prone versions
@@ -409,6 +410,7 @@ fn managed_output(node_id: u32, mode: Mode) -> VirtualOutput {
        ownership: DisplayOwnership::SessionManaged,
        reused_gen: None,
        pool_gen: None,
        expect_exact_dims: false,
    }
 }
@@ -506,7 +508,7 @@ pub fn launch_into_session(cmd: &str) -> Result<std::process::Child> {
    let mut c = Command::new("sh");
    c.arg("-c").arg(cmd);
    match discover_session_display_env() {
-        Some((x11, wayland)) => {
+        Some((x11, wayland, _xauth)) => {
            tracing::info!(
                command = %cmd,
                x11_display = x11.as_deref().unwrap_or("-"),
@@ -531,11 +533,71 @@ pub fn launch_into_session(cmd: &str) -> Result<std::process::Child> {
        .context("spawn launch command into gamescope session")
 }
-/// Find the live gamescope session's `(DISPLAY, WAYLAND_DISPLAY)` by scanning same-uid processes
+/// EVERY nested Xwayland the running gamescope session exposes, as `(DISPLAY, XAUTHORITY)` pairs
-/// for one whose environment carries `GAMESCOPE_WAYLAND_DISPLAY` (gamescope sets it for everything
+/// for the XFixes cursor source (remote-desktop-sweep Phase C). gamescope can run several
-/// it runs — Steam, the game, our own nested `sh`). The Wayland value returned is that gamescope
+/// (`--xwayland-count N` — Steam Gaming Mode uses 2: one for Big Picture, one for the game), and
-/// socket; `DISPLAY` is the nested Xwayland. Either can be individually absent.
+/// the pointer lives on whichever is FOCUSED — so the source connects to all and follows the one
-fn discover_session_display_env() -> Option<(Option<String>, Option<String>)> {
+/// whose pointer moves. The host is not a gamescope child, so gamescope's auth cookie rides along
 /// when a process exposes it. Empty when no gamescope session is running / none exposes a `DISPLAY`.
 #[cfg(target_os = "linux")]
 pub(crate) fn xwayland_cursor_targets() -> Vec<(String, Option<String>)> {
    // SAFETY: `getuid()` is a parameterless POSIX call that always succeeds and touches no memory.
    let uid = unsafe { libc::getuid() };
    let mut out: Vec<(String, Option<String>)> = Vec::new();
    let Ok(entries) = std::fs::read_dir("/proc") else {
        return out;
    };
    for e in entries.flatten() {
        let name = e.file_name();
        let Some(pid_str) = name.to_str() else {
            continue;
        };
        if !pid_str.bytes().all(|b| b.is_ascii_digit()) {
            continue;
        }
        let Ok(md) = std::fs::metadata(e.path()) else {
            continue;
        };
        use std::os::unix::fs::MetadataExt;
        if md.uid() != uid {
            continue;
        }
        let Ok(raw) = std::fs::read(e.path().join("environ")) else {
            continue;
        };
        let (mut display, mut is_gamescope, mut xauth) = (None, false, None);
        for kv in raw.split(|&b| b == 0) {
            let kv = String::from_utf8_lossy(kv);
            if kv.starts_with("GAMESCOPE_WAYLAND_DISPLAY=") {
                is_gamescope = true;
            } else if let Some(v) = kv.strip_prefix("DISPLAY=") {
                if !v.is_empty() {
                    display = Some(v.to_string());
                }
            } else if let Some(v) = kv.strip_prefix("XAUTHORITY=") {
                if !v.is_empty() {
                    xauth = Some(v.to_string());
                }
            }
        }
        if let (true, Some(d)) = (is_gamescope, display) {
            // Distinct DISPLAY only; prefer the first non-empty XAUTHORITY seen for it.
            match out.iter_mut().find(|(dd, _)| *dd == d) {
                Some((_, xa)) if xa.is_none() => *xa = xauth,
                Some(_) => {}
                None => out.push((d, xauth)),
            }
        }
    }
    out
 }
 /// Find the live gamescope session's `(DISPLAY, WAYLAND_DISPLAY, XAUTHORITY)` by scanning same-uid
 /// processes for one whose environment carries `GAMESCOPE_WAYLAND_DISPLAY` (gamescope sets it for
 /// everything it runs — Steam, the game, our own nested `sh`). The Wayland value returned is that
 /// gamescope socket; `DISPLAY` is the nested Xwayland; `XAUTHORITY` is its auth file (for X
 /// clients that aren't gamescope children). Any one can be individually absent.
 fn discover_session_display_env() -> Option<(Option<String>, Option<String>, Option<String>)> {
    // SAFETY: `getuid()` is a parameterless POSIX call that always succeeds and touches no memory.
    let uid = unsafe { libc::getuid() };
    for e in std::fs::read_dir("/proc").ok()?.flatten() {
@@ -558,6 +620,7 @@ fn discover_session_display_env() -> Option<(Option<String>, Option<String>)> {
        };
        let mut display = None;
        let mut gs_wayland = None;
        let mut xauth = None;
        for kv in raw.split(|&b| b == 0) {
            let kv = String::from_utf8_lossy(kv);
            if let Some(v) = kv.strip_prefix("GAMESCOPE_WAYLAND_DISPLAY=") {
@@ -568,11 +631,15 @@ fn discover_session_display_env() -> Option<(Option<String>, Option<String>)> {
                if !v.is_empty() {
                    display = Some(v.to_string());
                }
            } else if let Some(v) = kv.strip_prefix("XAUTHORITY=") {
                if !v.is_empty() {
                    xauth = Some(v.to_string());
                }
            }
        }
        // Only a process INSIDE a gamescope session (it has the marker var) is a valid source.
        if gs_wayland.is_some() {
-            return Some((display, gs_wayland));
+            return Some((display, gs_wayland, xauth));
        }
    }
    None
@@ -1116,6 +1183,26 @@ pub fn schedule_restore_tv_session() {
    }
 }
 /// Does any DRM connector report a physically `connected` display? Scans
 /// `/sys/class/drm/*/status` — only connector nodes (`card0-eDP-1`, `card0-HDMI-A-1`, …) have a
 /// `status` file, so the bare `cardN` device dirs and `renderD*` nodes filter themselves out. A
 /// headless box (VM, panel-less mini PC) has none — in which case a "restore to the physical
 /// panel" can only fail, gamescope having no output to drive. Errors (no DRM at all, sysfs
 /// unreadable) read as headless: the safe direction is keeping the working session.
 fn physical_display_connected() -> bool {
    connected_connector_under(std::path::Path::new("/sys/class/drm"))
 }
 /// [`physical_display_connected`] against an arbitrary sysfs root (the unit-testable core).
 fn connected_connector_under(base: &std::path::Path) -> bool {
    let Ok(entries) = std::fs::read_dir(base) else {
        return false;
    };
    entries.flatten().any(|e| {
        std::fs::read_to_string(e.path().join("status")).is_ok_and(|s| s.trim() == "connected")
    })
 }
 /// Tear down our host-managed session (freeing Steam) and restart the autologin gaming session(s)
 /// we stopped on connect — so the TV returns to gaming mode when no one is streaming. Invoked by
 /// [`start_restore_worker`] once the debounce deadline passes; takes the stopped-unit list so a
@@ -1127,6 +1214,19 @@ fn do_restore_tv_session() {
    {
        let mut took = STEAMOS_TOOK_OVER.lock().unwrap_or_else(|e| e.into_inner());
        if *took {
            // A box with no physically connected display (a VM, a panel-less mini PC) has no
            // "physical gaming session" to restore TO: removing the drop-in and restarting the
            // target just crash-loops gamescope (no output to drive) and strands every later
            // connect on "no usable compositor". Keep the headless session — and the takeover
            // state, so a same-mode reconnect reuses it warm — instead. Checked at restore time
            // (not connect time) so plugging a panel in later restores normally.
            if !physical_display_connected() {
                tracing::info!(
                    "gamescope (SteamOS): no physical display connected — keeping the headless \
                     session (nothing to restore to)"
                );
                return;
            }
            *took = false;
            clear_takeover(); // A3: takeover undone — drop the persisted crash-restore marker
            *MANAGED_SESSION.lock().unwrap_or_else(|e| e.into_inner()) = None;
@@ -1226,15 +1326,31 @@ pub fn start_restore_worker() -> std::sync::Arc<()> {
 /// session). Shared by the attach and host-managed-session paths.
 fn point_injector_at_eis() {
    match find_gamescope_eis_socket() {
-        Some(sock) => match std::fs::write(ei_socket_file(), &sock) {
+        Some(sock) => {
            // Relay format: line 1 = socket, optional line 2 = the session's CURRENT output
            // size as "WxH". gamescope's EIS advertises only a degenerate INT32_MAX region, so
            // the injector can't learn the output geometry from the protocol — the hint lets
            // it scale normalized client positions correctly even when the client streams at
            // a different resolution than the session runs (foreign attach, supersample).
            let size = current_gamescope_output_size();
            let body = match size {
                Some((w, h)) => format!("{sock}\n{w}x{h}"),
                None => sock.clone(),
            };
            match std::fs::write(ei_socket_file(), body) {
                Ok(()) => {
-                tracing::info!(socket = %sock, "gamescope: pointed injector at the session's EIS socket")
+                    tracing::info!(
                        socket = %sock,
                        output = ?size,
                        "gamescope: pointed injector at the session's EIS socket"
                    )
                }
                Err(e) => tracing::warn!(
                    error = %e,
                    "gamescope: could not write the EIS relay file — input may not reach the session"
                ),
-        },
+            }
        }
        None => tracing::warn!(
            "gamescope: no connectable gamescope EIS socket found — input won't reach the session"
        ),
@@ -1524,7 +1640,35 @@ impl Drop for GamescopeProc {
 #[cfg(test)]
 mod tests {
-    use super::{cgroup_is_punktfunk_owned, is_steam_launch, shape_dedicated_command};
+    use super::{
        cgroup_is_punktfunk_owned, connected_connector_under, is_steam_launch,
        shape_dedicated_command,
    };
    #[test]
    fn connector_status_scan() {
        let base = std::env::temp_dir().join(format!("pf-drm-scan-{}", std::process::id()));
        let mk = |name: &str, status: Option<&str>| {
            let dir = base.join(name);
            std::fs::create_dir_all(&dir).unwrap();
            if let Some(s) = status {
                std::fs::write(dir.join("status"), s).unwrap();
            }
        };
        // Headless layout: device + render nodes only (no status files) → not connected.
        mk("card0", None);
        mk("renderD128", None);
        assert!(!connected_connector_under(&base));
        // Connectors present but nothing plugged in → still not connected.
        mk("card0-HDMI-A-1", Some("disconnected\n"));
        assert!(!connected_connector_under(&base));
        // A live panel → connected.
        mk("card0-eDP-1", Some("connected\n"));
        assert!(connected_connector_under(&base));
        // A missing base dir (no DRM at all) reads as headless.
        assert!(!connected_connector_under(&base.join("nope")));
        std::fs::remove_dir_all(&base).unwrap();
    }
    #[test]
    fn steam_launch_detection() {
@@ -92,11 +92,19 @@ fn next_output_name() -> String {
 /// The Hyprland virtual-display driver. Stateless — each [`create`](VirtualDisplay::create) adds one
 /// named headless output and spins up a portal thread owning the cast on it.
-pub struct HyprlandDisplay;
+pub struct HyprlandDisplay {
    /// Out-of-band cursor request (`set_hw_cursor`, the negotiated cursor channel): portal
    /// `CursorMode::Metadata` — shapes/positions ride `SPA_META_Cursor` for the channel + the
    /// composite blend. Off (every non-channel session): `Embedded` — the compositor paints the
    /// pointer into frames, zero host-side cursor work (the pre-channel default this backend
    /// always had). ⚠️ Metadata is UNTESTED on-glass for this backend (Phase B wired it so the
    /// channel isn't silently dead here; KWin/Mutter are the validated legs).
    hw_cursor: bool,
 }
 impl HyprlandDisplay {
    pub fn new() -> Result<Self> {
-        Ok(HyprlandDisplay)
+        Ok(HyprlandDisplay { hw_cursor: false })
    }
 }
@@ -141,6 +149,14 @@ impl VirtualDisplay for HyprlandDisplay {
        "hyprland"
    }
    fn set_hw_cursor(&mut self, on: bool) {
        self.hw_cursor = on;
    }
    fn hw_cursor(&self) -> bool {
        self.hw_cursor
    }
    fn create(&mut self, mode: Mode) -> Result<VirtualOutput> {
        // Log the permission-system caveat once per process (silent black frames otherwise).
        preflight_once();
@@ -167,9 +183,10 @@ impl VirtualDisplay for HyprlandDisplay {
        let (setup_tx, setup_rx) = std::sync::mpsc::channel::<Result<(OwnedFd, u32), String>>();
        let stop = Arc::new(AtomicBool::new(false));
        let stop_thread = stop.clone();
        let hw_cursor = self.hw_cursor;
        thread::Builder::new()
            .name("punktfunk-hypr-vout".into())
-            .spawn(move || portal_thread(setup_tx, stop_thread))
+            .spawn(move || portal_thread(setup_tx, stop_thread, hw_cursor))
            .context("spawn hyprland portal thread")?;
        let (fd, node_id) = match setup_rx.recv_timeout(Duration::from_secs(20)) {
@@ -199,6 +216,7 @@ impl VirtualDisplay for HyprlandDisplay {
            ownership: DisplayOwnership::Owned,
            reused_gen: None,
            pool_gen: None,
            expect_exact_dims: false,
        })
    }
 }
@@ -490,7 +508,17 @@ fn ensure_xdph_config() -> Result<()> {
 /// stopped (the zbus connection is the cast's lifetime). xdph answers source selection via our
 /// custom picker, no dialog. (Kept separate from wlroots' copy so each wlr-family backend stays
 /// self-owned per D1; unify if they ever diverge no further.)
-fn portal_thread(setup_tx: Sender<Result<(OwnedFd, u32), String>>, stop: Arc<AtomicBool>) {
+fn portal_thread(
    setup_tx: Sender<Result<(OwnedFd, u32), String>>,
    stop: Arc<AtomicBool>,
    hw_cursor: bool,
 ) {
    // Portal cursor mode per the session's channel negotiation (see the struct doc).
    let cursor_mode = if hw_cursor {
        CursorMode::Metadata
    } else {
        CursorMode::Embedded
    };
    use ashpd::desktop::screencast::{CursorMode, Screencast, SelectSourcesOptions, SourceType};
    use ashpd::desktop::PersistMode;
    use ashpd::enumflags2::BitFlags;
@@ -523,7 +551,7 @@ fn portal_thread(setup_tx: Sender<Result<(OwnedFd, u32), String>>, stop: Arc<Ato
                .select_sources(
                    &session,
                    SelectSourcesOptions::default()
-                        .set_cursor_mode(CursorMode::Embedded)
+                        .set_cursor_mode(cursor_mode)
                        // xdph offers MONITOR; the custom picker selects our output.
                        .set_sources(BitFlags::from_flag(SourceType::Monitor))
                        .set_multiple(false)
@@ -56,8 +56,13 @@ use zkde::zkde_screencast_stream_unstable_v1::{
 };
 use zkde::zkde_screencast_unstable_v1::ZkdeScreencastUnstableV1 as Screencast;
-/// `pointer` attachment mode (the protocol enum): render the cursor into the stream so the
+/// `pointer` attachment modes (the protocol enum), chosen per session by `set_hw_cursor`
-/// remote sees it move with injected input.
+/// (Phase B — the Windows no-regression gate mirrored): a CURSOR-CHANNEL session gets METADATA
 /// (`SPA_META_Cursor` on the stream — shapes forwarded to the client, the composite flip blends
 /// host-side; embedded would leave both with nothing, the round-1 mutter trap), every other
 /// session gets EMBEDDED — KWin composites the pointer into frames itself, zero host-side
 /// cursor work, the pre-channel path Moonlight/legacy clients always had.
 const POINTER_METADATA: u32 = 4;
 const POINTER_EMBEDDED: u32 = 2;
 /// The name we give the created output; KWin exposes it to output-management as `Virtual-<name>`.
@@ -79,14 +84,19 @@ pub struct KwinDisplay {
    /// `None` for shared/anonymous) — reported to the registry via [`last_identity_slot`] so it can key
    /// the group arrangement + `/display/state` slot to the same id this backend named the output with.
    last_slot: Option<u32>,
-    /// The base output name the last `create` used (`punktfunk` / `punktfunk-<id>`) — so
+    /// The RESOLVED kscreen address of the last `create`'s output — the numeric kscreen output id
-    /// [`apply_position`](VirtualDisplay::apply_position) can address the KWin output `Virtual-<name>`.
+    /// when [`resolve_kscreen_addr`] found it, else the `Virtual-<name>` fallback — so
    /// [`apply_position`](VirtualDisplay::apply_position) addresses OUR output even while a
    /// superseded same-name sibling is still alive.
    last_name: Option<String>,
    /// The topology-restore action the last `create` prepared (re-enable the outputs an `exclusive`
    /// topology disabled), pending pickup by the registry via [`take_topology_restore`] — so the
    /// physical is re-enabled only when the display GROUP's last member drops (§6.1), not this session's.
    /// A backstop [`Drop`] runs it if the registry never took it (so a physical is never left dark).
    pending_restore: Option<Box<dyn FnOnce() + Send>>,
    /// Out-of-band cursor request (`set_hw_cursor`, i.e. the session negotiated the cursor
    /// channel): METADATA pointer mode at creation; off = EMBEDDED (see the consts above).
    hw_cursor: bool,
 }
 impl Drop for KwinDisplay {
@@ -123,12 +133,22 @@ impl VirtualDisplay for KwinDisplay {
        self.pending_restore.take()
    }
    fn set_hw_cursor(&mut self, on: bool) {
        self.hw_cursor = on;
    }
    fn hw_cursor(&self) -> bool {
        self.hw_cursor
    }
    fn apply_position(&mut self, x: i32, y: i32) {
-        let Some(name) = self.last_name.clone() else {
+        // `last_name` holds the RESOLVED kscreen address (numeric output id, or the
        // `Virtual-<name>` fallback) — never re-derive from the name: during a supersede two
        // outputs share it and the command would hit the old one (see `create`).
        let Some(output) = self.last_name.clone() else {
            return;
        };
-        let output = format!("Virtual-{name}");
+        // kscreen-doctor position syntax: `output.<name-or-id>.position.<x>,<y>`.
        // kscreen-doctor position syntax: `output.<name>.position.<x>,<y>`.
        let ok = std::process::Command::new("kscreen-doctor")
            .arg(format!("output.{output}.position.{x},{y}"))
            .status()
@@ -158,32 +178,100 @@ impl VirtualDisplay for KwinDisplay {
            None => VOUT_NAME.to_string(),
        };
        self.last_name = Some(name.clone()); // for apply_position (registry-driven §6.2 layout)
        let (width, height) = (mode.width, mode.height);
        let pointer_mode = if self.hw_cursor {
            POINTER_METADATA
        } else {
            POINTER_EMBEDDED
        };
        let spawn_vout = |w: u32, h: u32| -> Result<(u32, Arc<AtomicBool>)> {
            let (setup_tx, setup_rx) = std::sync::mpsc::channel::<Result<u32, String>>();
            let stop = Arc::new(AtomicBool::new(false));
            let stop_thread = stop.clone();
        let (width, height) = (mode.width, mode.height);
            let name_thread = name.clone();
            thread::Builder::new()
                .name("punktfunk-kwin-vout".into())
-            .spawn(move || virtual_output_thread(width, height, name_thread, setup_tx, stop_thread))
+                .spawn(move || {
                    virtual_output_thread(w, h, name_thread, pointer_mode, setup_tx, stop_thread)
                })
                .context("spawn KWin virtual-output thread")?;
-
+            match setup_rx.recv_timeout(Duration::from_secs(20)) {
-        let node_id = match setup_rx.recv_timeout(Duration::from_secs(20)) {
+                Ok(Ok(v)) => Ok((v, stop)),
            Ok(Ok(v)) => v,
                Ok(Err(e)) => bail!("KWin virtual output failed: {e}"),
                Err(_) => bail!("timed out creating the KWin virtual output"),
            }
        };
-        tracing::info!(node_id, width, height, "KWin virtual output ready");
+        // KWin creates virtual outputs at a hardcoded 60 Hz, `stream_virtual_output` has no
-        // KWin creates virtual outputs at a hardcoded 60 Hz and `stream_virtual_output` has no
+        // refresh argument — and its screencast stream builds its PipeWire format offer, INCLUDING
-        // refresh argument, so above 60 Hz we install + select a custom mode (supported on virtual
+        // the `maxFramerate` cap it actively throttles delivery to, ONCE at stream creation
-        // outputs since KWin 6.6) before capture connects PipeWire, so the stream negotiates at the
+        // (screencaststream.cpp `buildFormats`, verified against the live offer with `pw-dump`:
-        // higher rate. First cut shells out to kscreen-doctor; the in-process
+        // the offer stays `max=60` after a kscreen mode change, and consumers connecting later
-        // kde_output_management_v2 client is a follow-up. `set_custom_refresh` reads back and
+        // still negotiate against it). The ONLY path that rebuilds the offer is the stream's own
-        // returns what KWin *actually* achieved so the encoder paces to the real source rate (a
+        // resize handling: when the source's texture size changes while recording, KWin re-runs
-        // rejected custom mode leaves the output at 60 Hz). At ≤60 Hz there's nothing to install —
+        // `buildFormats` — picking up the output's CURRENT refresh — and renegotiates the live
-        // the source runs 60 Hz and the encoder downsamples — so carry the requested rate through.
+        // stream via `pw_stream_update_params`. So above 60 Hz the output is born at a
-        let achieved_hz = if mode.refresh_hz > 60 {
+        // SACRIFICIAL height: installing + selecting the real `WxH@hz` custom mode (supported on
-            set_custom_refresh(width, height, mode.refresh_hz, &name)
+        // virtual outputs since KWin 6.6) then changes the SIZE, and the first buffers recorded
        // after the consumer connects trigger KWin's resize → a renegotiation to `WxH@hz`. The
        // capturer holds frames until that lands (`expect_exact_dims`), so the pipeline never
        // builds against the birth mode. First cut shells out to kscreen-doctor; the in-process
        // kde_output_management_v2 client is a follow-up. `set_custom_refresh` reads back what
        // KWin *actually* gave so the encoder paces to the real source rate. At ≤60 Hz there's
        // nothing to install — the output is born at the real size and 60 Hz is the offer anyway.
        let want_high = mode.refresh_hz > 60;
        let birth_h = if want_high { height + 16 } else { height };
        let (mut node_id, mut stop) = spawn_vout(width, birth_h)?;
        tracing::info!(
            node_id,
            width,
            height,
            birth_h,
            embedded_pointer = !self.hw_cursor,
            "KWin virtual output ready"
        );
        // ⚠️ ADDRESS BY NUMERIC KSCREEN ID, NEVER BY NAME: a supersede (mode switch) creates the
        // replacement output — SAME per-slot name, deliberately, for KWin's per-name config
        // persistence — while the superseded sibling is still alive (create-before-drop). Every
        // name-addressed kscreen command then hits the FIRST match = the OLD output: on-glass this
        // resized the LIVE session's display out from under it (wrong-res/black), read back the
        // OLD output as "custom mode applied", made the OLD output primary, and positioned it —
        // while the new output never left its birth mode and the capturer's dims gate starved.
        // Resolve OUR output's kscreen id once (match: managed-prefix name AND current mode ==
        // the just-created birth size; newest id wins), and use it for every kscreen operation.
        let mut addr = resolve_kscreen_addr(&name, width, birth_h);
        self.last_name = Some(addr.clone()); // for apply_position (registry-driven §6.2 layout)
        let mut expect_exact_dims = false;
        let achieved_hz = if want_high {
            let (achieved, size_applied) =
                set_custom_refresh(width, height, mode.refresh_hz, &addr);
            if size_applied {
                // Real mode selected: the recording stream will renegotiate to it (see above).
                expect_exact_dims = true;
                achieved
            } else {
                // Custom-mode install/select rejected (pre-6.6 KWin / stale kscreen-doctor): the
                // output is STUCK at the sacrificial birth size — unusable. Recreate plain at the
                // real size (the pre-sacrifice behavior: correct size, KWin's native 60 Hz).
                tracing::warn!(
                    "KWin rejected the custom mode — recreating the virtual output at the real \
                     size (60 Hz ceiling on this KWin)"
                );
                stop.store(true, Ordering::Relaxed);
                // Let KWin retire the doomed output before re-using its name.
                std::thread::sleep(Duration::from_millis(300));
                let (nid, st) = spawn_vout(width, height)?;
                node_id = nid;
                stop = st;
                addr = resolve_kscreen_addr(&name, width, height);
                self.last_name = Some(addr.clone());
                tracing::info!(
                    node_id,
                    width,
                    height,
                    "KWin virtual output ready (fallback)"
                );
                60
            }
        } else {
            mode.refresh_hz
        };
@@ -194,9 +282,9 @@ impl VirtualDisplay for KwinDisplay {
        // bootstrap output. Read from the policy (replacing the PUNKTFUNK_KWIN_VIRTUAL_PRIMARY boolean).
        use crate::policy::Topology;
        let disabled = match crate::effective_topology() {
-            Topology::Exclusive => apply_virtual_primary(&name),
+            Topology::Exclusive => apply_virtual_primary(&addr),
            Topology::Primary => {
-                apply_virtual_primary_only(&name);
+                apply_virtual_primary_only(&addr);
                Vec::new() // nothing disabled → nothing to restore
            }
            Topology::Extend | Topology::Auto => Vec::new(),
@@ -212,11 +300,13 @@ impl VirtualDisplay for KwinDisplay {
        });
        // Layout position (§6.2) is applied by the registry via `apply_position` right after create
        // (it owns the display group, so it computes auto-row / manual placement over the whole group).
-        Ok(VirtualOutput::owned(
+        let mut out = VirtualOutput::owned(
            node_id,
            Some((mode.width, mode.height, achieved_hz)),
            Box::new(StopGuard { stop }),
-        ))
+        );
        out.expect_exact_dims = expect_exact_dims;
        Ok(out)
    }
 }
@@ -254,14 +344,88 @@ fn reenable_outputs(outputs: &[(String, String)]) {
    tracing::info!(reenabled = ?outputs, "KWin: restored the physical/bootstrap outputs at their captured modes (group empty)");
 }
-/// Best-effort: raise the just-created virtual output's refresh above KWin's default 60 Hz by
+/// Resolve the kscreen address of the virtual output the host JUST created: the managed-prefix
-/// installing + selecting a custom mode via `kscreen-doctor` (the output is `Virtual-<VOUT_NAME>`,
+/// name alone is ambiguous during a supersede (the replacement deliberately reuses the per-slot
-/// refresh given in mHz), then **read back the active mode** and return the refresh KWin actually
+/// name while the superseded sibling is still alive), so match on the birth mode's size too —
-/// gave us. The apply command can report success yet leave the output at 60 Hz (mode rejected),
+/// only the just-created output sits at the sacrificial `(w, h)` — and prefer the HIGHEST output
-/// and a silent rate mismatch surfaces downstream as judder / duplicated frames — so the caller
+/// id (the newest) if several match. Returns the numeric id as a string (kscreen-doctor accepts
-/// paces the encoder to the *achieved* rate, not the requested one.
+/// `output.<id>.…`), falling back to the ambiguous `Virtual-<name>` if the output hasn't reached
-fn set_custom_refresh(width: u32, height: u32, hz: u32, name: &str) -> u32 {
+/// kscreen's model yet after a few tries (single-output sessions are unambiguous anyway).
-    let output = format!("Virtual-{name}");
+fn resolve_kscreen_addr(name: &str, w: u32, h: u32) -> String {
    let fallback = format!("Virtual-{name}");
    for attempt in 0..3 {
        if attempt > 0 {
            std::thread::sleep(Duration::from_millis(150));
        }
        let Some(doc) = kscreen_json() else { continue };
        let Some(outputs) = doc.get("outputs").and_then(|o| o.as_array()) else {
            continue;
        };
        let best = outputs
            .iter()
            .filter(|o| {
                o.get("name")
                    .and_then(|n| n.as_str())
                    .is_some_and(|n| n.starts_with(&fallback))
                    && output_active_size(o) == Some((w, h))
            })
            .filter_map(|o| o.get("id").and_then(|i| i.as_u64()))
            .max();
        if let Some(id) = best {
            tracing::info!(id, name, w, h, "KWin: resolved the new output's kscreen id");
            return id.to_string();
        }
    }
    tracing::warn!(
        name,
        w,
        h,
        "KWin: could not resolve the new output's kscreen id — falling back to name addressing \
         (ambiguous during a mode-switch supersede)"
    );
    fallback
 }
 /// `kscreen-doctor -j` parsed, `None` on any failure.
 fn kscreen_json() -> Option<serde_json::Value> {
    let out = std::process::Command::new("kscreen-doctor")
        .arg("-j")
        .output()
        .ok()?;
    serde_json::from_slice(&out.stdout).ok()
 }
 /// The `(width, height)` of an output's CURRENT mode from its `kscreen-doctor -j` entry.
 fn output_active_size(o: &serde_json::Value) -> Option<(u32, u32)> {
    let as_id = |v: &serde_json::Value| -> Option<String> {
        v.as_str()
            .map(|s| s.to_string())
            .or_else(|| v.as_u64().map(|n| n.to_string()))
    };
    let current = o.get("currentModeId").and_then(as_id)?;
    let mode = o
        .get("modes")?
        .as_array()?
        .iter()
        .find(|m| m.get("id").and_then(as_id).as_deref() == Some(current.as_str()))?;
    let size = mode.get("size")?;
    Some((
        size.get("width").and_then(|v| v.as_u64())? as u32,
        size.get("height").and_then(|v| v.as_u64())? as u32,
    ))
 }
 /// Best-effort: install + select the `width`x`height`@`hz` custom mode on the just-created
 /// virtual output via `kscreen-doctor` (`output` is the RESOLVED kscreen address — numeric id or
 /// name, see [`resolve_kscreen_addr`] — refresh given in mHz), then **read back the active mode**
 /// and return `(achieved_hz, size_applied)`. The apply command can report success yet leave the
 /// output on its old mode (rejected), and a silent rate mismatch surfaces downstream as judder /
 /// duplicated frames — so the caller paces the encoder to the *achieved* rate, not the requested
 /// one. `size_applied` tells the sacrificial-birth caller (see `create`) whether the SIZE half of
 /// the mode actually landed — that, not the refresh, is what triggers KWin's stream
 /// renegotiation.
 fn set_custom_refresh(width: u32, height: u32, hz: u32, output: &str) -> (u32, bool) {
    let output = output.to_string();
    let mhz = hz.saturating_mul(1000);
    let run = |arg: String| {
        std::process::Command::new("kscreen-doctor")
@@ -275,26 +439,29 @@ fn set_custom_refresh(width: u32, height: u32, hz: u32, name: &str) -> u32 {
        "output.{output}.addCustomMode.{width}.{height}.{mhz}.full"
    ));
    let applied = run(format!("output.{output}.mode.{width}x{height}@{hz}"));
-    match read_active_refresh(&output) {
+    match read_active_mode(&output) {
-        Some(achieved) if achieved >= hz => {
+        Some((w, h, achieved)) => {
            let size_applied = (w, h) == (width, height);
            if achieved >= hz && size_applied {
                tracing::info!(
                    output,
                    requested = hz,
                    achieved,
                    "KWin virtual output: custom refresh applied"
                );
-            achieved
+            } else {
        }
        Some(achieved) => {
                tracing::warn!(
                    output,
                    requested = hz,
                    achieved,
                    active_w = w,
                    active_h = h,
                    applied,
-                "KWin virtual output refresh below requested — pacing the encoder to the achieved \
+                    "KWin virtual output mode below requested — pacing the encoder to the \
-                 rate (custom-mode install rejected? is kscreen-doctor up to date?)"
+                     achieved rate (custom-mode install rejected? is kscreen-doctor up to date?)"
                );
-            achieved.max(1)
+            }
            (achieved.max(1), size_applied)
        }
        None => {
            tracing::warn!(
@@ -304,38 +471,37 @@ fn set_custom_refresh(width: u32, height: u32, hz: u32, name: &str) -> u32 {
                "could not read back KWin virtual output refresh — assuming 60 Hz (is \
                 kscreen-doctor installed?)"
            );
-            60
+            (60, false)
        }
    }
 }
-/// Read the active refresh (Hz, rounded) of `output` from `kscreen-doctor -j`. `None` if the
+/// Read the active mode (`(width, height, refresh_hz)`, Hz rounded) of `output` — a RESOLVED
-/// tool, the output, or its current mode can't be found. Mode/output ids come through as either
+/// kscreen address (numeric id or name, see [`resolve_kscreen_addr`]) — from `kscreen-doctor -j`.
-/// JSON strings or numbers depending on the KWin version, so both are accepted.
+/// `None` if the tool, the output, or its current mode can't be found. Mode/output ids come
-fn read_active_refresh(output: &str) -> Option<u32> {
+/// through as either JSON strings or numbers depending on the KWin version, so both are accepted.
-    let out = std::process::Command::new("kscreen-doctor")
+fn read_active_mode(output: &str) -> Option<(u32, u32, u32)> {
-        .arg("-j")
+    let doc = kscreen_json()?;
        .output()
        .ok()?;
    let doc: serde_json::Value = serde_json::from_slice(&out.stdout).ok()?;
    let as_id = |v: &serde_json::Value| -> Option<String> {
        v.as_str()
            .map(|s| s.to_string())
            .or_else(|| v.as_u64().map(|n| n.to_string()))
    };
-    let o = doc
+    let o = doc.get("outputs")?.as_array()?.iter().find(|o| {
-        .get("outputs")?
+        o.get("name").and_then(|n| n.as_str()) == Some(output)
-        .as_array()?
+            || o.get("id").and_then(as_id).as_deref() == Some(output)
-        .iter()
+    })?;
        .find(|o| o.get("name").and_then(|n| n.as_str()) == Some(output))?;
    let current = o.get("currentModeId").and_then(as_id)?;
    let mode = o
        .get("modes")?
        .as_array()?
        .iter()
        .find(|m| m.get("id").and_then(as_id).as_deref() == Some(current.as_str()))?;
    let size = mode.get("size")?;
    let w = size.get("width").and_then(|v| v.as_u64())? as u32;
    let h = size.get("height").and_then(|v| v.as_u64())? as u32;
    let hz = mode.get("refreshRate").and_then(|r| r.as_f64())?;
-    Some(hz.round() as u32)
+    Some((w, h, hz.round() as u32))
 }
 /// The prefix EVERY managed KWin output shares — Stage 3 names them `punktfunk` / `punktfunk-<id>`,
@@ -374,7 +540,7 @@ fn output_current_mode_spec(o: &serde_json::Value) -> Option<String> {
 /// bare re-enable drops a 120 Hz panel to KWin's default ~60 Hz).
 /// **Group-aware (§6.1):** excludes the WHOLE managed family (the [`MANAGED_PREFIX`]), not just this
 /// session's own output — so a 2nd `exclusive` session (with a distinct per-slot name) never disables
-/// the 1st session's live output. Parsed from `kscreen-doctor -j` (same source as [`read_active_refresh`]).
+/// the 1st session's live output. Parsed from `kscreen-doctor -j` (same source as [`read_active_mode`]).
 fn other_enabled_outputs() -> Vec<(String, String)> {
    let out = match std::process::Command::new("kscreen-doctor")
        .arg("-j")
@@ -437,12 +603,12 @@ fn a_managed_output_is_primary() -> bool {
        .unwrap_or(false)
 }
-/// Set `Virtual-punktfunk` primary and disable the bootstrap output(s) so the managed group becomes
+/// Set our output primary and disable the bootstrap output(s) so the managed group becomes
-/// the sole desktop (KWin re-homes plasmashell + windows onto it). Returns the disabled outputs for
+/// the sole desktop (KWin re-homes plasmashell + windows onto it). `ours` is the RESOLVED kscreen
 /// address (numeric id or name, see [`resolve_kscreen_addr`]). Returns the disabled outputs for
 /// the keepalive to re-enable on teardown. Best-effort: on failure, streaming continues (just possibly
 /// showing only the wallpaper) rather than failing the session.
-fn apply_virtual_primary(name: &str) -> Vec<(String, String)> {
+fn apply_virtual_primary(ours: &str) -> Vec<(String, String)> {
    let ours = format!("Virtual-{name}");
    let kscreen = |args: &[String]| {
        std::process::Command::new("kscreen-doctor")
            .args(args)
@@ -480,8 +646,7 @@ fn apply_virtual_primary(name: &str) -> Vec<(String, String)> {
 /// **Primary** (Stage 2): make the streamed output the primary but KEEP the other outputs enabled
 /// (don't disable the bootstrap/physical) — so the shell re-homes onto the streamed surface while a
 /// physical screen stays usable. Nothing to restore on teardown (we disabled nothing).
-fn apply_virtual_primary_only(name: &str) {
+fn apply_virtual_primary_only(ours: &str) {
    let ours = format!("Virtual-{name}");
    let ok = std::process::Command::new("kscreen-doctor")
        .arg(format!("output.{ours}.primary"))
        .status()
@@ -578,10 +743,11 @@ fn virtual_output_thread(
    width: u32,
    height: u32,
    name: String,
    pointer_mode: u32,
    setup_tx: Sender<Result<u32, String>>,
    stop: Arc<AtomicBool>,
 ) {
-    if let Err(e) = run(width, height, &name, &setup_tx, &stop) {
+    if let Err(e) = run(width, height, &name, pointer_mode, &setup_tx, &stop) {
        // If we never delivered a node id, report the failure to the waiting opener.
        let _ = setup_tx.send(Err(format!("{e:#}")));
    }
@@ -622,6 +788,7 @@ fn run(
    width: u32,
    height: u32,
    name: &str,
    pointer_mode: u32,
    setup_tx: &Sender<Result<u32, String>>,
    stop: &AtomicBool,
 ) -> Result<()> {
@@ -642,13 +809,14 @@ fn run(
        )
    })?;
-    // Create the virtual output sized to the client, cursor composited into the stream.
+    // Create the virtual output sized to the client; the pointer rides as stream metadata
    // (cursor-channel session) or KWin embeds it into frames (everyone else — see the consts).
    let stream = screencast.stream_virtual_output(
        name.to_string(),
        width as i32,
        height as i32,
        1.0, // scale (logical == physical)
-        POINTER_EMBEDDED,
+        pointer_mode,
        &qh,
        (),
    );
@@ -46,7 +46,17 @@ const BUS_DC: &str = "org.gnome.Mutter.DisplayConfig";
 /// e.g. when our virtual output is torn down — so we never persist a layout to monitors.xml).
 const APPLY_TEMPORARY: u32 = 1;
-/// Mutter cursor mode: render the cursor into the stream (matches the KWin/gamescope backends).
+/// Mutter cursor mode: ship the pointer as `SPA_META_Cursor` metadata instead of burning it into
 /// the frames. The capturer always negotiates the meta (pf-capture `meta_param`) and the encoder
 /// blend composites it for sessions where the client does not draw the cursor itself — while a
 /// cursor-forwarding session strips the overlay and sends shape/state over the cursor channel.
 /// Embedded mode would leave BOTH paths blind: no metadata means nothing to forward AND nothing
 /// to blend, and Mutter's own embedded painting is what the pre-channel path relied on.
 const CURSOR_METADATA: u32 = 2;
 /// `cursor-mode` embedded (mutter enum 1): Mutter composites the pointer into frames itself —
 /// zero host-side cursor work, the pre-channel path. Chosen for every session WITHOUT the
 /// negotiated cursor channel (`set_hw_cursor` off — Phase B, the Windows no-regression gate
 /// mirrored); metadata stays the cursor-channel sessions' mode (shapes forwarded / host blend).
 const CURSOR_EMBEDDED: u32 = 1;
 /// Serializes, process-wide, every Mutter operation that adds/removes a virtual monitor or applies
@@ -69,6 +79,9 @@ pub struct MutterDisplay {
    /// sole-monitor config (which would disable the first session's virtual). Defaults true (a lone
    /// session establishes topology as before).
    first_in_group: bool,
    /// Out-of-band cursor request (`set_hw_cursor`): metadata cursor-mode at creation; off =
    /// embedded (see [`CURSOR_EMBEDDED`]).
    hw_cursor: bool,
    /// The connecting client's cert fingerprint (set before [`create`](VirtualDisplay::create)) —
    /// keys the per-client persisted **scale** (GNOME can't persist it itself: Mutter mints a fresh
    /// EDID serial per `RecordVirtual` monitor, so `monitors.xml` never rematches; see
@@ -84,6 +97,7 @@ impl MutterDisplay {
    pub fn new() -> Result<Self> {
        Ok(MutterDisplay {
            first_in_group: true,
            hw_cursor: false,
            client_fp: None,
            last_slot: None,
        })
@@ -112,6 +126,14 @@ impl VirtualDisplay for MutterDisplay {
        self.client_fp = fingerprint;
    }
    fn set_hw_cursor(&mut self, on: bool) {
        self.hw_cursor = on;
    }
    fn hw_cursor(&self) -> bool {
        self.hw_cursor
    }
    fn last_identity_slot(&self) -> Option<u32> {
        self.last_slot
    }
@@ -141,6 +163,7 @@ impl VirtualDisplay for MutterDisplay {
        let stop = Arc::new(AtomicBool::new(false));
        let stop_thread = stop.clone();
        let first_in_group = self.first_in_group;
        let hw_cursor = self.hw_cursor;
        thread::Builder::new()
            .name("punktfunk-mutter-vout".into())
            .spawn(move || {
@@ -149,6 +172,7 @@ impl VirtualDisplay for MutterDisplay {
                    stop_thread,
                    mode,
                    first_in_group,
                    hw_cursor,
                    scale_key,
                    remembered_scale,
                )
@@ -203,6 +227,7 @@ fn session_thread(
    stop: Arc<AtomicBool>,
    mode: Mode,
    first_in_group: bool,
    hw_cursor: bool,
    scale_key: String,
    remembered_scale: Option<f64>,
 ) {
@@ -262,7 +287,7 @@ fn session_thread(
            }
        };
-        let session = match connect(mode, remembered_scale).await {
+        let session = match connect(mode, hw_cursor, remembered_scale).await {
            Ok(s) => s,
            Err(e) => {
                let _ = setup_tx.send(Err(format!("{e:#}")));
@@ -363,7 +388,11 @@ struct MutterSession {
 /// desktop scale, passed as the virtual mode's `preferred-scale` so Mutter creates the monitor
 /// already scaled (Mutter ≥ 48; older Mutter ignores unknown mode keys) — this covers the
 /// `extend` topology, where we never issue our own ApplyMonitorsConfig.
-async fn connect(mode: Mode, preferred_scale: Option<f64>) -> Result<MutterSession> {
+async fn connect(
    mode: Mode,
    hw_cursor: bool,
    preferred_scale: Option<f64>,
 ) -> Result<MutterSession> {
    let conn = zbus::Connection::session()
        .await
        .context("connect session D-Bus")?;
@@ -424,7 +453,14 @@ async fn connect(mode: Mode, preferred_scale: Option<f64>) -> Result<MutterSessi
    // once gated behind PUNKTFUNK_MUTTER_VIRTUAL_REFRESH; the stop-screencast-before-any-monitor-
    // reconfig teardown below fixed the crash, so pinning the client's refresh is now the default.)
    let mut rec: HashMap<&str, Value> = HashMap::new();
-    rec.insert("cursor-mode", Value::from(CURSOR_EMBEDDED));
+    rec.insert(
        "cursor-mode",
        Value::from(if hw_cursor {
            CURSOR_METADATA
        } else {
            CURSOR_EMBEDDED
        }),
    );
    if mode.refresh_hz > 60 || preferred_scale.is_some() {
        let mut vmode: HashMap<&str, Value> = HashMap::new();
        vmode.insert("size", Value::from((mode.width, mode.height)));
@@ -54,11 +54,19 @@ chooser_cmd=cat {} 2>/dev/null || echo 'Monitor: HEADLESS-1'\n",
 /// The wlroots/Sway virtual-display driver. Stateless — each [`create`](VirtualDisplay::create)
 /// adds one headless output and spins up a portal thread owning the cast on it.
-pub struct WlrootsDisplay;
+pub struct WlrootsDisplay {
    /// Out-of-band cursor request (`set_hw_cursor`, the negotiated cursor channel): portal
    /// `CursorMode::Metadata` — shapes/positions ride `SPA_META_Cursor` for the channel + the
    /// composite blend. Off (every non-channel session): `Embedded` — the compositor paints the
    /// pointer into frames, zero host-side cursor work (the pre-channel default this backend
    /// always had). ⚠️ Metadata is UNTESTED on-glass for this backend (Phase B wired it so the
    /// channel isn't silently dead here; KWin/Mutter are the validated legs).
    hw_cursor: bool,
 }
 impl WlrootsDisplay {
    pub fn new() -> Result<Self> {
-        Ok(WlrootsDisplay)
+        Ok(WlrootsDisplay { hw_cursor: false })
    }
 }
@@ -73,6 +81,14 @@ impl VirtualDisplay for WlrootsDisplay {
        "wlroots"
    }
    fn set_hw_cursor(&mut self, on: bool) {
        self.hw_cursor = on;
    }
    fn hw_cursor(&self) -> bool {
        self.hw_cursor
    }
    fn create(&mut self, mode: Mode) -> Result<VirtualOutput> {
        let before = output_names()
            .context("swaymsg get_outputs (is the host inside the sway session env — SWAYSOCK?)")?;
@@ -104,9 +120,10 @@ impl VirtualDisplay for WlrootsDisplay {
        let (setup_tx, setup_rx) = std::sync::mpsc::channel::<Result<(OwnedFd, u32), String>>();
        let stop = Arc::new(AtomicBool::new(false));
        let stop_thread = stop.clone();
        let hw_cursor = self.hw_cursor;
        thread::Builder::new()
            .name("punktfunk-wlr-vout".into())
-            .spawn(move || portal_thread(setup_tx, stop_thread))
+            .spawn(move || portal_thread(setup_tx, stop_thread, hw_cursor))
            .context("spawn wlroots portal thread")?;
        let (fd, node_id) = match setup_rx.recv_timeout(Duration::from_secs(20)) {
@@ -136,6 +153,7 @@ impl VirtualDisplay for WlrootsDisplay {
            ownership: DisplayOwnership::Owned,
            reused_gen: None,
            pool_gen: None,
            expect_exact_dims: false,
        })
    }
 }
@@ -254,7 +272,17 @@ fn ensure_xdpw_config() -> Result<()> {
 /// The ScreenCast portal handshake (same shape as the capture module's portal thread, but it
 /// reports the fd + node id and parks until stopped — the zbus connection is the cast's
 /// lifetime). xdpw answers the source selection via the chooser, no dialog.
-fn portal_thread(setup_tx: Sender<Result<(OwnedFd, u32), String>>, stop: Arc<AtomicBool>) {
+fn portal_thread(
    setup_tx: Sender<Result<(OwnedFd, u32), String>>,
    stop: Arc<AtomicBool>,
    hw_cursor: bool,
 ) {
    // Portal cursor mode per the session's channel negotiation (see the struct doc).
    let cursor_mode = if hw_cursor {
        CursorMode::Metadata
    } else {
        CursorMode::Embedded
    };
    use ashpd::desktop::screencast::{CursorMode, Screencast, SelectSourcesOptions, SourceType};
    use ashpd::desktop::PersistMode;
    use ashpd::enumflags2::BitFlags;
@@ -287,7 +315,7 @@ fn portal_thread(setup_tx: Sender<Result<(OwnedFd, u32), String>>, stop: Arc<Ato
                .select_sources(
                    &session,
                    SelectSourcesOptions::default()
-                        .set_cursor_mode(CursorMode::Embedded)
+                        .set_cursor_mode(cursor_mode)
                        // xdpw offers MONITOR only; the chooser picks our output.
                        .set_sources(BitFlags::from_flag(SourceType::Monitor))
                        .set_multiple(false)
@@ -84,20 +84,28 @@ fn topology_str() -> String {
 /// `quit` is the session's deliberate-quit flag: when the session ends with it set (the client closed
 /// with the quit application code — a user "stop", not a network drop), the display is torn down
 /// **immediately**, skipping the keep-alive linger. A bare disconnect leaves it `false` → normal linger.
 ///
 /// `supersedes`: the pool gen of a display this acquire REPLACES (a mid-stream mode switch creates
 /// the new display before retiring the old — create-before-drop). The replacement inherits group
 /// topology ownership: without this, the dying predecessor counts as a live sibling and the new
 /// display "extends" behind it, losing a Primary/Exclusive topology on every resize. `None`
 /// everywhere else.
 pub fn acquire(
    vd: &mut Box<dyn super::VirtualDisplay>,
    mode: super::Mode,
    quit: std::sync::Arc<std::sync::atomic::AtomicBool>,
    supersedes: Option<u64>,
 ) -> Result<super::VirtualOutput> {
    let backend = vd.name();
    #[cfg(target_os = "linux")]
-    let out = linux::acquire(vd, mode, quit);
+    let out = linux::acquire(vd, mode, quit, supersedes);
    #[cfg(not(target_os = "linux"))]
    let out = {
        // Windows leases in the manager (its own linger); its deliberate-quit skip is wired through
        // `VirtualDisplay::set_quit_flag` on the backend instance (set by the session before any
-        // `create`, so the retry-hold lease gets it too) — not through this parameter.
+        // `create`, so the retry-hold lease gets it too) — not through this parameter. The
-        let _ = quit;
+        // supersede handoff is Linux-pool-only too (the manager resizes in place).
        let _ = (quit, supersedes);
        vd.create(mode)
    };
    if out.is_ok() {
@@ -260,6 +268,11 @@ mod linux {
        /// Generation stamp: a [`DisplayLease`] only releases if its gen still matches (a stale lease
        /// — its entry was reused + re-stamped — is a no-op).
        gen: u64,
        /// The out-of-band-cursor mode this display was CREATED with (Phase B): metadata-pointer
        /// (cursor-channel session) vs compositor-embedded. Reuse requires an exact match — a kept
        /// embedded display has no cursor metadata for a channel session to forward, and a kept
        /// metadata display would leave a channel-less session with no pointer in its frames.
        hw_cursor: bool,
    }
    /// A per-group topology-restore action (see [`Entry::topology_restore`]).
@@ -421,6 +434,7 @@ mod linux {
        vd: &mut Box<dyn VirtualDisplay>,
        mode: Mode,
        quit: Arc<AtomicBool>,
        supersedes: Option<u64>,
    ) -> Result<VirtualOutput> {
        ensure_timer();
        let backend = vd.name();
@@ -464,6 +478,7 @@ mod linux {
                        ) && e.backend == backend
                            && e.mode == mode
                            && e.launch == launch
                            && e.hw_cursor == vd.hw_cursor()
                            && epoch_matches(e.backend, e.epoch, cur_epoch)
                    })
                    .map(|e| (e.gen, e.node_id))
@@ -535,9 +550,19 @@ mod linux {
        // §6.1) — so a topology-establishing backend (Mutter exclusive) extends into an already-exclusive
        // desktop rather than re-clobbering the first session's virtual. Best-effort (a concurrent create
        // is a narrow race); single-session is always `first == true` → today's behavior.
        // Siblings that don't demote the newcomer:
        //   * the display this acquire SUPERSEDES (mode switch, create-before-drop) — still Active
        //     here because the old lease drops only after the new pipeline is up, but it's leaving,
        //     and deferring to it loses the group's Primary/Exclusive topology on every resize;
        //   * kept (Lingering/Pinned) entries — no session owns them, so there is no live desktop to
        //     clobber; a new session next to an unclaimed leftover should still establish topology.
        let first_in_group = {
            let es = r.entries.lock().unwrap();
-            !es.iter().any(|e| e.backend == backend)
+            !es.iter().any(|e| {
                e.backend == backend
                    && Some(e.gen) != supersedes
                    && matches!(e.life, lifecycle::State::Active { .. })
            })
        };
        vd.set_first_in_group(first_in_group);
@@ -565,6 +590,11 @@ mod linux {
        let node_id = real.node_id;
        let preferred_mode = real.preferred_mode;
        // Fresh creates only: the backend may have birthed the output at a sacrificial mode whose
        // stream must renegotiate before frames count (KWin >60 Hz — see backend.rs). A REUSED kept
        // display already renegotiated in its prior session (the producer's rebuilt offer persists
        // across consumer reconnects), so the reuse path above correctly leaves the flag off.
        let expect_exact_dims = real.expect_exact_dims;
        // The backend's topology-restore action (KWin `exclusive` → re-enable the disabled physicals),
        // lifted into the group so it runs once when the group's last member drops (§6.1), not at this
        // session's teardown. `None` for non-exclusive / non-first / backends whose topology auto-reverts.
@@ -584,6 +614,7 @@ mod linux {
            launch: launch.clone(),
            epoch: cur_epoch,
            gen,
            hw_cursor: vd.hw_cursor(),
        };
        // Compute this new display's position in its group (design §6.2) BEFORE pushing, then push
@@ -628,7 +659,9 @@ mod linux {
        if (position.x, position.y) != (0, 0) {
            vd.apply_position(position.x, position.y);
        }
-        Ok(output_for(node_id, preferred_mode, gen, quit, false))
+        let mut out = output_for(node_id, preferred_mode, gen, quit, false);
        out.expect_exact_dims = expect_exact_dims;
        Ok(out)
    }
    /// The linger a releasing session actually gets. A deliberate quit (`force_immediate` — the
@@ -1003,6 +1036,7 @@ mod linux {
                launch: None,
                epoch: 0,
                gen,
                hw_cursor: false,
            }
        }
@@ -162,6 +162,15 @@ pub fn launch_into_gamescope_session(cmd: &str) -> Result<std::process::Child> {
    gamescope::launch_into_session(cmd)
 }
 /// Every nested Xwayland `(DISPLAY, XAUTHORITY)` of the running gamescope session for the XFixes
 /// cursor source (remote-desktop-sweep Phase C) — gamescope can run several, and the pointer is on
 /// whichever is focused. Empty when no gamescope session is running / it exposes no Xwayland (the
 /// host then leaves gamescope cursorless, today's behaviour).
 #[cfg(target_os = "linux")]
 pub fn gamescope_xwayland_cursor_targets() -> Vec<(String, Option<String>)> {
    gamescope::xwayland_cursor_targets()
 }
 /// B2: has a **dedicated** gamescope game session's game exited (its `node_id` doesn't reappear within a
 /// short window after capture loss)? The dedicated-spawn session ends cleanly on `true` instead of the
 /// capture-loss rebuild. Scoped to the session's OWN node so a coexisting gamescope doesn't mask the
@@ -207,6 +216,20 @@ pub fn cancel_pending_tv_restore() {
 #[cfg(not(target_os = "linux"))]
 pub fn cancel_pending_tv_restore() {}
 /// Can the MANAGED gamescope path stand a session up from nothing on this box (SteamOS's
 /// `gamescope-session` launcher or Bazzite's `gamescope-session-plus` present)? Lets the connect
 /// path route a "no live graphical session" box to the gamescope takeover — which rebuilds the
 /// session at the client's mode — instead of failing the connect. Always `false` off Linux.
 #[cfg(target_os = "linux")]
 pub fn managed_session_available() -> bool {
    gamescope::managed_session_available()
 }
 #[cfg(not(target_os = "linux"))]
 pub fn managed_session_available() -> bool {
    false
 }
 /// Call when a client session ends: if the host-managed gamescope path took over a box's autologin
 /// gaming session (stopped its single-instance Steam to stream at the client's mode), **schedule** a
 /// debounced restore so the TV returns to gaming mode — unless a client reconnects within the window
@@ -71,6 +71,15 @@ struct Monitor {
    /// 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>,
    /// This monitor was ADDed with the v5 hardware-cursor flag (already driver-proto-gated) —
    /// preserved across a re-arrival resize so the recreated monitor keeps the cursor channel.
    hw_cursor: bool,
    /// The ADD reply flagged this monitor's OS target as carrying an IRREVOCABLE hardware-cursor
    /// declare from an earlier session (§8.6): DWM excludes the pointer from its frames forever
    /// (the sticky state survives monitor REMOVE→ADD via the stable per-client target id), so a
    /// session WITHOUT the cursor channel must self-composite — carried into
    /// [`WinCaptureTarget`] for the IDD-push capturer's forced-composite gate.
    cursor_excluded: bool,
    /// 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. The SAME
    /// process for every parallel monitor (one devnode → one WUDFHost hosts all publishers), which
@@ -98,6 +107,7 @@ impl Monitor {
                gdi_name: n,
                target_id: self.target_id,
                wudf_pid: self.wudf_pid,
                cursor_excluded: self.cursor_excluded,
            })
    }
 }
@@ -267,10 +277,44 @@ pub fn vdm() -> &'static VirtualDisplayManager {
 /// for the process lifetime — a dead one is RETIRED (kept alive, see [`DeviceSlot`]), so a stale copy
 /// can only fail IOCTLs, never dangle. `None` before the first backend open — impossible for a
 /// capturer, which only exists on a monitor the manager created.
 /// Can this host's pf-vdisplay driver run the v5 hardware-cursor channel? Reads the
 /// handshake-latched protocol version, opening the control device once if no session has
 /// opened it yet this service run (the same open every session performs anyway) — so the
 /// Welcome-time capability decision never guesses. `false` when the driver is missing/stale.
 ///
 /// The FIRST session's Welcome precedes any backend construction (`vdisplay::open` runs at
 /// display prep, after the Welcome), so this must not assume an initialised manager —
 /// `init` is idempotent and constructing the driver facade is free (on-glass finding: the
 /// `vdm()` expect panicked the very first handshake of a fresh service).
 pub fn hw_cursor_capable() -> bool {
    let m = init(Box::new(crate::driver::PfVdisplayDriver));
    let v = m.driver_proto.load(Ordering::Relaxed);
    if v != 0 {
        return v >= 5;
    }
    let _ = m.ensure_device();
    m.driver_proto.load(Ordering::Relaxed) >= 5
 }
 pub fn control_device_handle() -> Option<HANDLE> {
    VDM.get().and_then(VirtualDisplayManager::device_handle)
 }
 /// Re-commit the CURRENT display config under the manager `state` lock (the sole-topology-mutator
 /// contract of [`force_mode_reenumeration`]). The secure-desktop guard's actuator: the OS only
 /// reverts a path to its software-cursor default ON a mode commit, so standing the hardware-cursor
 /// declare down (`IOCTL_SET_CURSOR_FORWARD` off) needs this nudge for UAC/Winlogon to actually
 /// render. `false` (no-op) before the first backend open — no monitors exist to re-commit for.
 pub fn force_recommit() -> bool {
    let Some(m) = VDM.get() else {
        return false;
    };
    let _guard = m.state.lock().unwrap();
    // SAFETY: `force_mode_reenumeration`'s contract is "call under the manager `state` lock";
    // held above. The call reads + re-applies the current CCD config over owned locals.
    unsafe { pf_win_display::win_display::force_mode_reenumeration() }
 }
 /// Best-effort "is this WUDFHost pid still alive?" — the monitor-liveness probe for the JOIN path.
 /// `OpenProcess` failing (pid reaped) or the process being signaled ⇒ dead. Pid reuse could
 /// theoretically alias a fresh process and read "alive"; the joining session then just retries into
@@ -390,6 +434,7 @@ impl VirtualDisplayManager {
        mode: Mode,
        client_fp: Option<[u8; 32]>,
        client_hdr: Option<punktfunk_core::quic::HdrMeta>,
        hw_cursor: bool,
        quit: Option<Arc<AtomicBool>>,
    ) -> Result<VirtualOutput> {
        // Console-session guard: a host outside the ACTIVE console session cannot drive the display
@@ -622,7 +667,9 @@ impl VirtualDisplayManager {
        // SAFETY: `create_monitor` requires `dev` to be a valid control handle; `dev` is the handle
        // `ensure_device()` returned above (cached handles are never closed — a dead one is retired,
        // kept alive; see `DeviceSlot`), and we hold the `state` lock.
-        let mon = match unsafe { self.create_monitor(dev, mode, slot, client_hdr, &mut inner) } {
+        let mon = match unsafe {
            self.create_monitor(dev, mode, slot, client_hdr, hw_cursor, &mut inner)
        } {
            // The cached device died under us (driver upgrade / WUDFHost restart, detected only
            // now — e.g. the host sat idle past the pinger-less window). Retire it, reopen, and
            // retry ONCE so the reconnect-after-driver-restart succeeds first try instead of
@@ -635,7 +682,7 @@ impl VirtualDisplayManager {
                );
                // SAFETY: as above — `dev` is the handle the reopening `ensure_device` just
                // returned, and the `state` lock is still held.
-                unsafe { self.create_monitor(dev, mode, slot, client_hdr, &mut inner)? }
+                unsafe { self.create_monitor(dev, mode, slot, client_hdr, hw_cursor, &mut inner)? }
            }
            r => r?,
        };
@@ -861,6 +908,7 @@ impl VirtualDisplayManager {
        mode: Mode,
        slot: u32,
        client_hdr: Option<punktfunk_core::quic::HdrMeta>,
        hw_cursor: bool,
        inner: &mut MgrInner,
    ) -> Result<Monitor> {
        // The slot id doubles as the driver-preferred monitor id (EDID serial / ConnectorIndex), so
@@ -868,13 +916,17 @@ impl VirtualDisplayManager {
        // `0` (anonymous) = the driver auto-allocates the lowest-free id.
        let preferred_id = slot;
        let render_pin = resolve_render_pin();
        // Hardware cursor only against a driver that implements the v5 channel: an older driver
        // ignores the AddRequest field anyway (composited cursor), but gating here keeps the
        // capture layer from creating + delivering a section nobody will ever publish into.
        let hw_cursor = hw_cursor && self.driver_proto.load(Ordering::Relaxed) >= 5;
        // 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.
        // `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, render_pin, preferred_id, client_hdr)?
+                .add_monitor(dev, mode, render_pin, preferred_id, client_hdr, hw_cursor)?
        };
        // Mandatory keepalive: ping inside the watchdog window or the driver tears all displays down.
@@ -1061,6 +1113,8 @@ impl VirtualDisplayManager {
            resolved_monitor_id: added.resolved_monitor_id,
            position: (0, 0),
            gen: self.gen.fetch_add(1, Ordering::Relaxed),
            hw_cursor,
            cursor_excluded: added.cursor_excluded,
        })
    }
@@ -1231,7 +1285,7 @@ impl VirtualDisplayManager {
        // values passed by value — no borrow crosses the call.
        let added = unsafe {
            self.driver
-                .add_monitor(dev, mode, render_pin, slot, client_hdr)
+                .add_monitor(dev, mode, render_pin, slot, client_hdr, old.hw_cursor)
                .context("re-arrival ADD at the new mode")?
        };
        self.ensure_pinger();
@@ -1283,6 +1337,10 @@ impl VirtualDisplayManager {
            resolved_monitor_id: added.resolved_monitor_id,
            position: old.position,
            gen: old.gen,
            hw_cursor: old.hw_cursor,
            // Fresh from THIS reply, not `old`: the driver's per-target declare registry is the
            // ground truth (this session may itself have declared since the original ADD).
            cursor_excluded: added.cursor_excluded,
        })
    }
@@ -22,6 +22,11 @@ pub(crate) struct AddedMonitor {
    pub luid: LUID,
    pub wudf_pid: u32,
    pub resolved_monitor_id: u32,
    /// The driver reports the OS target already carries an IRREVOCABLE hardware-cursor declare
    /// from an earlier session (`AddReply::cursor_excluded`, remote-desktop-sweep §8.6): DWM
    /// excludes the pointer from this target's frames forever, so a session without the cursor
    /// channel must self-composite (GDI poller + blend) or stream a cursor-less desktop.
    pub cursor_excluded: bool,
 }
 /// The backend-specific IOCTL surface — the *only* thing that differs between SudoVDA and pf-vdisplay.
@@ -56,6 +61,7 @@ pub(crate) trait VdisplayDriver: Send + Sync {
        render_luid: Option<LUID>,
        preferred_monitor_id: u32,
        client_hdr: Option<punktfunk_core::quic::HdrMeta>,
        hw_cursor: bool,
    ) -> Result<AddedMonitor>;
    /// Refresh the LIVE monitor `key`'s advertised mode list to lead with `mode` (the in-place
    /// mid-stream resize, latency plan P2 — pf-vdisplay `IOCTL_UPDATE_MODES`, driver protocol v4).
@@ -234,6 +234,55 @@ pub unsafe fn send_frame_channel(dev: HANDLE, req: &control::SetFrameChannelRequ
    .context("pf-vdisplay SET_FRAME_CHANNEL")
 }
 /// Deliver a monitor's hardware-cursor section (`IOCTL_SET_CURSOR_CHANNEL`, proto v5) — the
 /// cursor sibling of [`send_frame_channel`], same delivery/ownership contract.
 ///
 /// # Safety
 /// `dev` must be a live pf-vdisplay control handle (see [`super::manager::control_device_handle`]).
 pub unsafe fn send_cursor_channel(
    dev: HANDLE,
    req: &control::SetCursorChannelRequest,
 ) -> Result<()> {
    let mut none: [u8; 0] = [];
    // SAFETY: per this fn's contract `dev` is the live control handle; `bytes_of(req)` borrows the
    // caller's request across this synchronous call; no output buffer.
    unsafe {
        ioctl(
            dev,
            control::IOCTL_SET_CURSOR_CHANNEL,
            bytemuck::bytes_of(req),
            &mut none,
        )
    }
    .map(|_| ())
    .context("pf-vdisplay SET_CURSOR_CHANNEL")
 }
 /// Flip a LIVE monitor's hardware-cursor declaration (`IOCTL_SET_CURSOR_FORWARD`, proto v6) —
 /// the mid-stream cursor-render flip. Fails against a pre-v6 driver (unknown IOCTL); callers
 /// log and keep the declared-at-ADD behavior.
 ///
 /// # Safety
 /// `dev` must be a live pf-vdisplay control handle (see [`super::manager::control_device_handle`]).
 pub unsafe fn send_cursor_forward(
    dev: HANDLE,
    req: &control::SetCursorForwardRequest,
 ) -> Result<()> {
    let mut none: [u8; 0] = [];
    // SAFETY: per this fn's contract `dev` is the live control handle; `bytes_of(req)` borrows the
    // caller's request across this synchronous call; no output buffer.
    unsafe {
        ioctl(
            dev,
            control::IOCTL_SET_CURSOR_FORWARD,
            bytemuck::bytes_of(req),
            &mut none,
        )
    }
    .map(|_| ())
    .context("pf-vdisplay SET_CURSOR_FORWARD")
 }
 /// RAII over a SetupAPI device-info list: every exit path of [`open_device`] destroys it (the error
 /// paths used to leak one `HDEVINFO` per failed open — and a driverless / mid-upgrade box probes
 /// repeatedly).
@@ -460,6 +509,7 @@ impl VdisplayDriver for PfVdisplayDriver {
        render_luid: Option<LUID>,
        preferred_monitor_id: u32,
        client_hdr: Option<punktfunk_core::quic::HdrMeta>,
        hw_cursor: bool,
    ) -> Result<AddedMonitor> {
        let session_id = next_session_id();
        // The client display's volume rides into the monitor's EDID CTA HDR block; all-zero =
@@ -485,7 +535,11 @@ impl VdisplayDriver for PfVdisplayDriver {
            max_luminance_nits,
            max_frame_avg_nits,
            min_luminance_millinits,
-            _reserved: 0,
+            // v5 cursor channel: the driver declares an IddCx hardware cursor for this monitor
            // (DWM stops compositing the pointer into the frame); the capture layer delivers the
            // CursorShm section right after its ring. Zero toward older drivers is harmless —
            // the host only sets this when the handshake-reported proto is >= 5.
            hw_cursor: hw_cursor as u32,
        };
        // SET_RENDER_ADAPTER (opt-in; pf-vdisplay IMPLEMENTS it). Non-fatal on failure: the driver reports
        // its real render LUID in the shared header, so the host binds correctly even if this is ignored.
@@ -551,10 +605,13 @@ impl VdisplayDriver for PfVdisplayDriver {
        })?;
        // Fail closed on a short reply — `target_id`/`wudf_pid`/`luid` below feed OpenProcess + the
        // WUDFHost verification, so don't decode a partially-written (zeroed) reply as authoritative.
-        if (n as usize) < size_of::<control::AddReply>() {
+        // The LEGACY size, not the full struct: an un-upgraded driver writes only the prefix before
        // the `cursor_excluded` tail; `out` is zero-initialized, so the missing tail reads `0`
        // (= unknown/clean — exactly what a driver that can't track declares should report).
        if (n as usize) < control::ADD_REPLY_LEGACY_SIZE {
            anyhow::bail!(
-                "pf-vdisplay ADD returned {n} bytes, expected {}",
+                "pf-vdisplay ADD returned {n} bytes, expected at least {}",
-                size_of::<control::AddReply>()
+                control::ADD_REPLY_LEGACY_SIZE
            );
        }
        // `pod_read_unaligned` (NOT `from_bytes`): `out` is a stack `[u8; N]` with no guaranteed 4-byte
@@ -569,6 +626,7 @@ impl VdisplayDriver for PfVdisplayDriver {
            target_id = reply.target_id,
            adapter_luid = %format_args!("{:#x}", luid.LowPart),
            wudf_pid = reply.wudf_pid,
            cursor_excluded = reply.cursor_excluded != 0,
            "pf-vdisplay monitor created {}x{}@{}",
            mode.width,
            mode.height,
@@ -605,6 +663,7 @@ impl VdisplayDriver for PfVdisplayDriver {
            luid,
            wudf_pid: reply.wudf_pid,
            resolved_monitor_id: reply.resolved_monitor_id,
            cursor_excluded: reply.cursor_excluded != 0,
        })
    }
@@ -682,6 +741,10 @@ pub struct PfVdisplayDisplay {
    /// freshly created monitor's EDID advertises this volume so host apps tone-map to the client's
    /// real panel.
    client_hdr: Option<punktfunk_core::quic::HdrMeta>,
    /// Declare an IddCx hardware cursor on the created monitor (the M2c cursor channel). Set by
    /// [`set_hw_cursor`](VirtualDisplay::set_hw_cursor) before `create`; only honored when the
    /// driver handshake reported proto >= 5.
    hw_cursor: bool,
    /// The session's deliberate-quit flag (`None` = no signal → the linger policy applies). Set by
    /// [`set_quit_flag`](VirtualDisplay::set_quit_flag) before `create`; rides into every lease this
    /// backend mints so a user "stop" tears the monitor down immediately instead of lingering.
@@ -694,6 +757,7 @@ impl PfVdisplayDisplay {
        Ok(Self {
            client_fp: None,
            client_hdr: None,
            hw_cursor: false,
            quit: None,
        })
    }
@@ -712,12 +776,26 @@ impl VirtualDisplay for PfVdisplayDisplay {
        self.client_hdr = hdr;
    }
    fn set_hw_cursor(&mut self, on: bool) {
        self.hw_cursor = on;
    }
    fn hw_cursor(&self) -> bool {
        self.hw_cursor
    }
    fn set_quit_flag(&mut self, quit: std::sync::Arc<std::sync::atomic::AtomicBool>) {
        self.quit = Some(quit);
    }
    fn create(&mut self, mode: Mode) -> Result<VirtualOutput> {
-        super::manager::vdm().acquire(mode, self.client_fp, self.client_hdr, self.quit.clone())
+        super::manager::vdm().acquire(
            mode,
            self.client_fp,
            self.client_hdr,
            self.hw_cursor,
            self.quit.clone(),
        )
    }
 }
@@ -804,17 +882,12 @@ mod tests {
        // Live-run diagnostics: surface the manager/backend tracing (activation ladder, settle
        // waits, UPDATE_MODES) on stdout — a bare test harness has no subscriber, which made the
        // first on-glass run blind.
-        let _ = tracing_subscriber::fmt()
+        // (tracing-subscriber is not a dep of this crate — run the host binary for traced runs.)
            .with_env_filter(
                tracing_subscriber::EnvFilter::try_from_default_env()
                    .unwrap_or_else(|_| "debug".into()),
            )
            .try_init();
        // Context probe: can this process see the CCD active-path set at all? (`None` = the query
        // itself fails in this session/window-station — the whole ladder would be blind, and a
        // "monitor never activated" verdict would be an artifact of the test context.)
        // SAFETY: CCD query over an owned empty slice (test-only diagnostics).
-        let active0 = unsafe { crate::win_display::count_other_active(&[]) };
+        let active0 = unsafe { pf_win_display::win_display::count_other_active(&[]) };
        println!("spike: CCD active paths visible before create: {active0:?}");
        let mut vd = PfVdisplayDisplay::new().expect("open pf-vdisplay");
        let first = vd
@@ -847,7 +920,7 @@ mod tests {
            .target_id;
        let in_place = t1 == t2;
        // SAFETY: CCD query over a Copy target id (test-only diagnostics).
-        let active = unsafe { crate::win_display::active_resolution(t2) };
+        let active = unsafe { pf_win_display::win_display::active_resolution(t2) };
        println!(
            "in-place resize spike: in_place={in_place} (target {t1} -> {t2}) took {resize_ms} ms, \
             active resolution now {active:?}"
@@ -23,10 +23,10 @@ use windows::core::PCWSTR;
 use windows::Win32::Devices::Display::{
    DisplayConfigGetDeviceInfo, DisplayConfigSetDeviceInfo, GetDisplayConfigBufferSizes,
    QueryDisplayConfig, SetDisplayConfig, DISPLAYCONFIG_DEVICE_INFO_GET_ADVANCED_COLOR_INFO,
-    DISPLAYCONFIG_DEVICE_INFO_GET_SOURCE_NAME, DISPLAYCONFIG_DEVICE_INFO_GET_TARGET_NAME,
+    DISPLAYCONFIG_DEVICE_INFO_GET_SDR_WHITE_LEVEL, DISPLAYCONFIG_DEVICE_INFO_GET_SOURCE_NAME,
-    DISPLAYCONFIG_DEVICE_INFO_SET_ADVANCED_COLOR_STATE, DISPLAYCONFIG_GET_ADVANCED_COLOR_INFO,
+    DISPLAYCONFIG_DEVICE_INFO_GET_TARGET_NAME, DISPLAYCONFIG_DEVICE_INFO_SET_ADVANCED_COLOR_STATE,
-    DISPLAYCONFIG_MODE_INFO, DISPLAYCONFIG_MODE_INFO_TYPE_SOURCE,
+    DISPLAYCONFIG_GET_ADVANCED_COLOR_INFO, DISPLAYCONFIG_MODE_INFO,
-    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_COMPONENT_VIDEO,
+    DISPLAYCONFIG_MODE_INFO_TYPE_SOURCE, DISPLAYCONFIG_OUTPUT_TECHNOLOGY_COMPONENT_VIDEO,
    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_COMPOSITE_VIDEO,
    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_DISPLAYPORT_EMBEDDED,
    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_DISPLAYPORT_EXTERNAL, DISPLAYCONFIG_OUTPUT_TECHNOLOGY_DVI,
@@ -35,10 +35,11 @@ use windows::Win32::Devices::Display::{
    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_SDI, DISPLAYCONFIG_OUTPUT_TECHNOLOGY_SDTVDONGLE,
    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_SVIDEO, DISPLAYCONFIG_OUTPUT_TECHNOLOGY_UDI_EMBEDDED,
    DISPLAYCONFIG_OUTPUT_TECHNOLOGY_UDI_EXTERNAL, DISPLAYCONFIG_PATH_INFO,
-    DISPLAYCONFIG_SET_ADVANCED_COLOR_STATE, DISPLAYCONFIG_SOURCE_DEVICE_NAME,
+    DISPLAYCONFIG_SDR_WHITE_LEVEL, DISPLAYCONFIG_SET_ADVANCED_COLOR_STATE,
-    DISPLAYCONFIG_TARGET_DEVICE_NAME, DISPLAYCONFIG_VIDEO_OUTPUT_TECHNOLOGY, QDC_ALL_PATHS,
+    DISPLAYCONFIG_SOURCE_DEVICE_NAME, DISPLAYCONFIG_TARGET_DEVICE_NAME,
-    QDC_ONLY_ACTIVE_PATHS, SDC_ALLOW_CHANGES, SDC_APPLY, SDC_FORCE_MODE_ENUMERATION,
+    DISPLAYCONFIG_VIDEO_OUTPUT_TECHNOLOGY, QDC_ALL_PATHS, QDC_ONLY_ACTIVE_PATHS, SDC_ALLOW_CHANGES,
-    SDC_SAVE_TO_DATABASE, SDC_TOPOLOGY_EXTEND, SDC_USE_SUPPLIED_DISPLAY_CONFIG,
+    SDC_APPLY, SDC_FORCE_MODE_ENUMERATION, SDC_SAVE_TO_DATABASE, SDC_TOPOLOGY_EXTEND,
    SDC_USE_SUPPLIED_DISPLAY_CONFIG,
 };
 use windows::Win32::Foundation::POINTL;
 use windows::Win32::Graphics::Gdi::{
@@ -481,6 +482,52 @@ pub unsafe fn advanced_color_enabled(target_id: u32) -> Option<bool> {
    None
 }
 /// The target's SDR white level as a SCALE relative to 80 nits (`1.0` = 80 nits): where DWM
 /// places SDR-white when composing SDR content onto this HDR desktop. An SDR-authored overlay
 /// (the composited cursor) must be multiplied by this in scRGB space or it renders visibly
 /// darker than the surrounding SDR desktop content (the Windows "SDR content brightness"
 /// slider default alone is ~2.5x). `None` = query failed / target not active (callers keep
 /// their last value or 1.0).
 ///
 /// # Safety
 /// Runs the read-only CCD query FFI over owned locals (same shape as [`advanced_color_enabled`]).
 pub unsafe fn sdr_white_level_scale(target_id: u32) -> Option<f32> {
    let mut np = 0u32;
    let mut nm = 0u32;
    if GetDisplayConfigBufferSizes(QDC_ONLY_ACTIVE_PATHS, &mut np, &mut nm).is_err() {
        return None;
    }
    let mut paths = vec![DISPLAYCONFIG_PATH_INFO::default(); np as usize];
    let mut modes = vec![DISPLAYCONFIG_MODE_INFO::default(); nm as usize];
    if QueryDisplayConfig(
        QDC_ONLY_ACTIVE_PATHS,
        &mut np,
        paths.as_mut_ptr(),
        &mut nm,
        modes.as_mut_ptr(),
        None,
    )
    .is_err()
    {
        return None;
    }
    for p in paths.iter().take(np as usize) {
        if p.targetInfo.id == target_id {
            let mut info = DISPLAYCONFIG_SDR_WHITE_LEVEL::default();
            info.header.r#type = DISPLAYCONFIG_DEVICE_INFO_GET_SDR_WHITE_LEVEL;
            info.header.size = size_of::<DISPLAYCONFIG_SDR_WHITE_LEVEL>() as u32;
            info.header.adapterId = p.targetInfo.adapterId;
            info.header.id = p.targetInfo.id;
            if DisplayConfigGetDeviceInfo(&mut info.header) == 0 && info.SDRWhiteLevel > 0 {
                // Contract: SDRWhiteLevel/1000 * 80 = nits, i.e. the /1000 IS the 80-nit scale.
                return Some(info.SDRWhiteLevel as f32 / 1000.0);
            }
            return None;
        }
    }
    None
 }
 /// Force the freshly-added virtual monitor to the client's exact `WxH@Hz`. The ADD IOCTL only
 /// ADVERTISES the mode; Windows otherwise activates an IDD target at a 1280x720 default, so the
 /// ACTIVE mode (what DXGI Desktop Duplication captures) must be set explicitly. CDS_TEST first so a
@@ -861,7 +908,18 @@ pub unsafe fn isolate_displays_ccd(keep_target_ids: &[u32]) -> Option<SavedConfi
                continue;
            }
            if p.flags & DISPLAYCONFIG_PATH_ACTIVE != 0 {
-                p.flags &= !DISPLAYCONFIG_PATH_ACTIVE; // mark this path inactive
+                // Mark the path inactive AND unpin its modes: per the SetDisplayConfig
                // contract a path being turned OFF needs BOTH mode indexes marked invalid,
                // and leaving them referencing the queried mode entries gets the whole
                // supplied config rejected with 0x57 ERROR_INVALID_PARAMETER on some
                // driver/topology combinations (field-reported: exclusive mode left the
                // physical panel lit, every retry failing 0x57). Writing the all-ones
                // sentinel to the whole union is also correct under the virtual-mode-aware
                // interpretation (cloneGroupId/sourceModeInfoIdx both become their 0xffff
                // INVALID values).
                p.flags &= !DISPLAYCONFIG_PATH_ACTIVE;
                p.sourceInfo.Anonymous.modeInfoIdx = DISPLAYCONFIG_PATH_MODE_IDX_INVALID;
                p.targetInfo.Anonymous.modeInfoIdx = DISPLAYCONFIG_PATH_MODE_IDX_INVALID;
                others += 1;
            }
        }
@@ -870,6 +928,26 @@ pub unsafe fn isolate_displays_ccd(keep_target_ids: &[u32]) -> Option<SavedConfi
        // never calls ASSIGN_SWAPCHAIN, so the driver receives no frames. SDC_FORCE_MODE_ENUMERATION
        // forces the re-commit; SAVE_TO_DATABASE only in the sole-path case (matches prior behavior —
        // don't permanently rewrite the user's multi-display layout; the teardown restore handles it).
        let rc = if others > 0 && attempt >= 2 {
            // ESCALATION (attempt 2+): supply ONLY the keep paths. Field-reported (AMD +
            // pf-vdisplay): carrying the doomed path in the array — inactive, modes unpinned —
            // gets the whole config rejected 0x57 on EVERY retry, so the loop alone never
            // converged; the same host applies the keep-only shape rc=0 whenever the topology
            // database has already made the virtual display sole. The final attempt also drops
            // SDC_FORCE_MODE_ENUMERATION in case the driver rejects it combined with a real
            // topology change — an actual path removal drives COMMIT_MODES on its own, so the
            // re-commit rationale doesn't need the flag here.
            let (kp, km) = keep_only_supplied(&paths, &modes);
            let mut esc = SDC_APPLY | SDC_USE_SUPPLIED_DISPLAY_CONFIG | SDC_ALLOW_CHANGES;
            if attempt < 4 {
                esc |= SDC_FORCE_MODE_ENUMERATION;
            }
            tracing::info!(
                "display isolate (CCD): escalating to a keep-only supplied config (attempt {attempt}/4, paths {}→{}, modes {}→{})",
                paths.len(), kp.len(), modes.len(), km.len()
            );
            SetDisplayConfig(Some(kp.as_slice()), Some(km.as_slice()), esc)
        } else {
            let mut flags = SDC_APPLY
                | SDC_USE_SUPPLIED_DISPLAY_CONFIG
                | SDC_ALLOW_CHANGES
@@ -877,7 +955,8 @@ pub unsafe fn isolate_displays_ccd(keep_target_ids: &[u32]) -> Option<SavedConfi
            if others == 0 {
                flags |= SDC_SAVE_TO_DATABASE;
            }
-        let rc = SetDisplayConfig(Some(paths.as_slice()), Some(modes.as_slice()), flags);
+            SetDisplayConfig(Some(paths.as_slice()), Some(modes.as_slice()), flags)
        };
        // A failed apply must be VISIBLE even when the verification below passes vacuously (nothing
        // else was active to deactivate — the lid-closed laptop case, where the success INFO used to
        // swallow rc=0x5): the re-commit above is load-bearing (COMMIT_MODES → ASSIGN_SWAPCHAIN),
@@ -904,6 +983,63 @@ pub unsafe fn isolate_displays_ccd(keep_target_ids: &[u32]) -> Option<SavedConfi
    Some(saved)
 }
 /// Build the ESCALATED supplied config for [`isolate_displays_ccd`]: ONLY the paths still flagged
 /// ACTIVE (the keep set — the caller already cleared ACTIVE on every doomed path), with the mode
 /// table rebuilt to just the entries those paths reference (indexes remapped). Docs-wise the
 /// dropped inactive entries were declared ignored anyway ("Only the paths within this array that
 /// have the DISPLAYCONFIG_PATH_ACTIVE flag set are set"), so this shape asks for the identical
 /// topology — minus the array contents some driver/OS validation combos reject with 0x57.
 unsafe fn keep_only_supplied(
    paths: &[DISPLAYCONFIG_PATH_INFO],
    modes: &[DISPLAYCONFIG_MODE_INFO],
 ) -> (Vec<DISPLAYCONFIG_PATH_INFO>, Vec<DISPLAYCONFIG_MODE_INFO>) {
    let mut out_paths = Vec::new();
    let mut out_modes = Vec::new();
    // old mode index → new. Shared entries dedup through here: a clone-style pair references ONE
    // source mode, and the docs require each source/target mode to appear in the table only once.
    let mut remap = std::collections::HashMap::new();
    for p in paths {
        if p.flags & DISPLAYCONFIG_PATH_ACTIVE == 0 {
            continue;
        }
        let mut q = *p;
        q.sourceInfo.Anonymous.modeInfoIdx = remap_mode_idx(
            q.sourceInfo.Anonymous.modeInfoIdx,
            modes,
            &mut out_modes,
            &mut remap,
        );
        q.targetInfo.Anonymous.modeInfoIdx = remap_mode_idx(
            q.targetInfo.Anonymous.modeInfoIdx,
            modes,
            &mut out_modes,
            &mut remap,
        );
        out_paths.push(q);
    }
    (out_paths, out_modes)
 }
 /// Move `modes[old]` into `out` (once — `remap` dedups) and return its new index. INVALID and
 /// out-of-range indexes stay INVALID — `SDC_ALLOW_CHANGES` lets best-mode logic fill the gap.
 fn remap_mode_idx(
    old: u32,
    modes: &[DISPLAYCONFIG_MODE_INFO],
    out: &mut Vec<DISPLAYCONFIG_MODE_INFO>,
    remap: &mut std::collections::HashMap<u32, u32>,
 ) -> u32 {
    if old == DISPLAYCONFIG_PATH_MODE_IDX_INVALID {
        return old;
    }
    let Some(m) = modes.get(old as usize) else {
        return DISPLAYCONFIG_PATH_MODE_IDX_INVALID;
    };
    *remap.entry(old).or_insert_with(|| {
        out.push(*m);
        (out.len() - 1) as u32
    })
 }
 /// The desktop-space rectangle `(x, y, w, h)` of `target_id`'s SOURCE — where this display's
 /// region lives in the desktop coordinate space. `None` while the target isn't an active path.
 /// Used by the IDD-push compose kick to dirty THE TARGET display: with parallel displays the
@@ -7,8 +7,12 @@
 //!   and ffmpeg's `hevc_nvenc` (encode thread) — each thread makes it current before use;
 //! * device memory: pitched allocations, the reusable `BufferPool`/`DeviceBuffer`, IPC
 //!   export/import, host readback, and the plane copies;
-//! * GL / external-memory interop (`RegisteredTexture`, `ExternalDmabuf`); and
+//! * GL / external-memory interop (`RegisteredTexture`, `ExternalDmabuf`).
-//! * the CUDA cursor-blend kernel (`CursorBlend`).
+//!
 //! (The CUDA cursor-blend PTX kernel that used to live here is retired: vendored PTX is JIT'd
 //! against the driver's ISA ceiling and silently dies on older drivers. The NVENC cursor blend
 //! is now the SPIR-V compute pass in [`super::vkslot`], dispatched over Vulkan-allocated,
 //! CUDA-imported input slots.)
 //!
 //! (We use GL interop, not EGL interop: `cuGraphicsEGLRegisterImage` is Tegra-only on the desktop
 //! driver — see [`super::egl`].)
@@ -18,7 +22,6 @@
 #![deny(clippy::undocumented_unsafe_blocks)]
 use anyhow::{bail, Result};
 use std::ffi::CStr;
 use std::os::raw::{c_uint, c_void};
 use std::sync::{Arc, Mutex, OnceLock};
@@ -280,258 +283,6 @@ pub fn copy_stream_handle() -> *mut c_void {
 /// Max cursor-overlay bitmap edge (px) uploaded to the device blend buffer — matches the Vulkan path.
 pub const CURSOR_MAX: u32 = 256;
 /// GPU cursor-overlay compositor for the NVENC path (cursor-as-metadata): loads the `cursor_blend`
 /// PTX module once and blends a straight-alpha RGBA cursor into an encoder-OWNED NVENC input surface
 /// (ARGB / NV12 / YUV444) with a small kernel launched over the cursor's rectangle — no full-frame
 /// pass, and the compositor's dmabuf is never touched. The cursor bitmap lives in a device buffer
 /// re-uploaded only when it changes. Requires `context()` to have succeeded (driver present).
 pub struct CursorBlend {
    module: CUmodule,
    f_argb: CUfunction,
    f_nv12: CUfunction,
    f_yuv444: CUfunction,
    cur_buf: CUdeviceptr, // device RGBA staging (CURSOR_MAX²·4, tight rows)
 }
 // SAFETY: process-lifetime driver handles used only from the encode thread with the shared context
 // current — like [`DeviceBuffer`], moving the struct between threads cannot dangle or race.
 unsafe impl Send for CursorBlend {}
 impl CursorBlend {
    /// Load the embedded PTX image and resolve the three blend kernels + a device cursor buffer.
    pub fn new(ptx: &[u8]) -> Result<CursorBlend> {
        // cuModuleLoadData reads a PTX image as a NUL-terminated string; the embedded .ptx is not,
        // so append a terminator.
        let mut image = ptx.to_vec();
        image.push(0);
        let mut module: CUmodule = std::ptr::null_mut();
        // SAFETY: `&mut module` is a live out-param the driver fills; `image` is a NUL-terminated PTX
        // byte image that outlives the synchronous load. `ck` bails on error before `module` is used.
        unsafe {
            ck(
                cuModuleLoadData(&mut module, image.as_ptr() as *const c_void),
                "cuModuleLoadData(cursor_blend)",
            )?;
        }
        let getf = |name: &CStr| -> Result<CUfunction> {
            let mut f: CUfunction = std::ptr::null_mut();
            // SAFETY: `module` loaded above; each name is a valid NUL-terminated symbol present in
            // the module (verified in the .ptx `.entry` list); `&mut f` is a live out-param.
            unsafe {
                ck(
                    cuModuleGetFunction(&mut f, module, name.as_ptr()),
                    "cuModuleGetFunction",
                )?;
            }
            Ok(f)
        };
        let f_argb = getf(c"blend_argb")?;
        let f_nv12 = getf(c"blend_nv12")?;
        let f_yuv444 = getf(c"blend_yuv444")?;
        let mut cur_buf: CUdeviceptr = 0;
        // SAFETY: `&mut cur_buf` is a live out-param; the size fits the CURSOR_MAX² RGBA buffer.
        unsafe {
            ck(
                cuMemAlloc_v2(&mut cur_buf, (CURSOR_MAX * CURSOR_MAX * 4) as usize),
                "cuMemAlloc(cursor)",
            )?;
        }
        Ok(CursorBlend {
            module,
            f_argb,
            f_nv12,
            f_yuv444,
            cur_buf,
        })
    }
    /// Upload the cursor RGBA (`cw*ch*4`, tight rows) into the device blend buffer. Call only when
    /// the bitmap changes; position moves are just kernel args.
    pub fn upload(&self, rgba: &[u8], cw: u32, ch: u32) -> Result<()> {
        let cw = cw.min(CURSOR_MAX);
        let ch = ch.min(CURSOR_MAX);
        let row = cw as usize * 4;
        let copy = CUDA_MEMCPY2D {
            srcMemoryType: 1, // HOST
            srcHost: rgba.as_ptr() as *const c_void,
            srcPitch: row,
            dstMemoryType: CU_MEMORYTYPE_DEVICE,
            dstDevice: self.cur_buf,
            dstPitch: row,
            WidthInBytes: row,
            Height: ch as usize,
            ..Default::default()
        };
        // SAFETY: HOST→DEVICE 2D copy of `row*ch` bytes; `rgba` covers at least that (caller passes
        // `cw*ch*4`), `cur_buf` is the CURSOR_MAX²·4 device alloc (row ≤ CURSOR_MAX·4, ch ≤ CURSOR_MAX).
        // Synchronous via `copy_blocking`. Requires the context current (caller's contract).
        unsafe { copy_blocking(&copy, "cursor HtoD") }
    }
    /// Blend into a packed 4-byte (NVENC ARGB) owned surface at `(ox,oy)`.
    #[allow(clippy::too_many_arguments)] // surface geometry + cursor size + offset — a struct would just be unpacked at the call
    pub fn blend_argb(
        &self,
        surf: CUdeviceptr,
        pitch: usize,
        w: u32,
        h: u32,
        cw: u32,
        ch: u32,
        ox: i32,
        oy: i32,
        sync: bool,
    ) -> Result<()> {
        let (mut a_surf, mut a_cur) = (surf, self.cur_buf);
        let (mut a_pitch, mut a_w, mut a_h) = (pitch as i32, w as i32, h as i32);
        let (mut a_cw, mut a_ch) = (cw.min(CURSOR_MAX) as i32, ch.min(CURSOR_MAX) as i32);
        let (mut a_ox, mut a_oy) = (ox, oy);
        let mut args: [*mut c_void; 9] = [
            &mut a_surf as *mut _ as *mut c_void,
            &mut a_pitch as *mut _ as *mut c_void,
            &mut a_w as *mut _ as *mut c_void,
            &mut a_h as *mut _ as *mut c_void,
            &mut a_cur as *mut _ as *mut c_void,
            &mut a_cw as *mut _ as *mut c_void,
            &mut a_ch as *mut _ as *mut c_void,
            &mut a_ox as *mut _ as *mut c_void,
            &mut a_oy as *mut _ as *mut c_void,
        ];
        self.launch(self.f_argb, a_cw as u32, a_ch as u32, &mut args, sync)
    }
    /// Blend into an owned planar YUV444 surface (3 stacked full-res planes) at `(ox,oy)`.
    #[allow(clippy::too_many_arguments)] // surface geometry + cursor size + offset — a struct would just be unpacked at the call
    pub fn blend_yuv444(
        &self,
        base: CUdeviceptr,
        pitch: usize,
        w: u32,
        h: u32,
        cw: u32,
        ch: u32,
        ox: i32,
        oy: i32,
        sync: bool,
    ) -> Result<()> {
        let (mut a_base, mut a_cur) = (base, self.cur_buf);
        let (mut a_pitch, mut a_w, mut a_h) = (pitch as i32, w as i32, h as i32);
        let (mut a_cw, mut a_ch) = (cw.min(CURSOR_MAX) as i32, ch.min(CURSOR_MAX) as i32);
        let (mut a_ox, mut a_oy) = (ox, oy);
        let mut args: [*mut c_void; 9] = [
            &mut a_base as *mut _ as *mut c_void,
            &mut a_pitch as *mut _ as *mut c_void,
            &mut a_w as *mut _ as *mut c_void,
            &mut a_h as *mut _ as *mut c_void,
            &mut a_cur as *mut _ as *mut c_void,
            &mut a_cw as *mut _ as *mut c_void,
            &mut a_ch as *mut _ as *mut c_void,
            &mut a_ox as *mut _ as *mut c_void,
            &mut a_oy as *mut _ as *mut c_void,
        ];
        self.launch(self.f_yuv444, a_cw as u32, a_ch as u32, &mut args, sync)
    }
    /// Blend into an owned NV12 surface (Y plane at `base`, interleaved UV at `base + pitch*h`).
    #[allow(clippy::too_many_arguments)] // surface geometry + cursor size + offset — a struct would just be unpacked at the call
    pub fn blend_nv12(
        &self,
        base: CUdeviceptr,
        pitch: usize,
        w: u32,
        h: u32,
        cw: u32,
        ch: u32,
        ox: i32,
        oy: i32,
        sync: bool,
    ) -> Result<()> {
        let (mut a_yb, mut a_uvb, mut a_cur) = (base, base + pitch as u64 * h as u64, self.cur_buf);
        let (mut a_yp, mut a_uvp) = (pitch as i32, pitch as i32);
        let (mut a_w, mut a_h) = (w as i32, h as i32);
        let (mut a_cw, mut a_ch) = (cw.min(CURSOR_MAX) as i32, ch.min(CURSOR_MAX) as i32);
        let (mut a_ox, mut a_oy) = (ox, oy);
        let mut args: [*mut c_void; 11] = [
            &mut a_yb as *mut _ as *mut c_void,
            &mut a_yp as *mut _ as *mut c_void,
            &mut a_uvb as *mut _ as *mut c_void,
            &mut a_uvp as *mut _ as *mut c_void,
            &mut a_w as *mut _ as *mut c_void,
            &mut a_h as *mut _ as *mut c_void,
            &mut a_cur as *mut _ as *mut c_void,
            &mut a_cw as *mut _ as *mut c_void,
            &mut a_ch as *mut _ as *mut c_void,
            &mut a_ox as *mut _ as *mut c_void,
            &mut a_oy as *mut _ as *mut c_void,
        ];
        // One thread per 2x2 luma block → grid over ceil(cw/2) × ceil(ch/2).
        self.launch(
            self.f_nv12,
            (a_cw as u32).div_ceil(2),
            (a_ch as u32).div_ceil(2),
            &mut args,
            sync,
        )
    }
    /// Launch `f` over a `work_w × work_h` grid (16×16 blocks) on the copy stream; `sync` waits
    /// for it, `!sync` leaves completion to the stream (stream-ordered consumers only — the
    /// kernel PARAMETERS are copied at launch time, so the arg locals need not outlive the call).
    fn launch(
        &self,
        f: CUfunction,
        work_w: u32,
        work_h: u32,
        args: &mut [*mut c_void],
        sync: bool,
    ) -> Result<()> {
        if work_w == 0 || work_h == 0 {
            return Ok(());
        }
        const B: u32 = 16;
        let stream = copy_stream();
        // SAFETY: `f` is a resolved kernel from our loaded module; `args` holds pointers to live
        // locals whose types match the kernel's C parameters (per the call site above) — CUDA
        // copies the parameter values during `cuLaunchKernel` itself, so they need not outlive
        // the call. Grid/block dims are non-zero. Launched on the copy stream (ordered after the
        // input-surface copy issued on the same stream); `sync` waits, `!sync` leaves ordering to
        // the stream (the NVENC IO-stream binding). Requires the context current.
        unsafe {
            ck(
                cuLaunchKernel(
                    f,
                    work_w.div_ceil(B),
                    work_h.div_ceil(B),
                    1,
                    B,
                    B,
                    1,
                    0,
                    stream,
                    args.as_mut_ptr(),
                    std::ptr::null_mut(),
                ),
                "cuLaunchKernel(cursor)",
            )?;
            if sync {
                ck(cuStreamSynchronize(stream), "cuStreamSynchronize(cursor)")?;
            }
            Ok(())
        }
    }
 }
 impl Drop for CursorBlend {
    fn drop(&mut self) {
        // SAFETY: `cur_buf`/`module` are our own handles, freed exactly once here; the context is
        // current on the encode thread that drops the encoder. Errors are ignored on teardown.
        unsafe {
            let _ = cuMemFree_v2(self.cur_buf);
            let _ = cuModuleUnload(self.module);
        }
    }
 }
 /// Allocate one pitched device buffer for `width`x`height` 4-byte pixels; returns `(ptr, pitch)`.
 fn alloc_pitched(width: u32, height: u32) -> Result<(CUdeviceptr, usize)> {
    let mut ptr: CUdeviceptr = 0;
@@ -9,7 +9,7 @@
 #![deny(clippy::undocumented_unsafe_blocks)]
 use anyhow::{bail, Result};
-use std::os::raw::{c_char, c_int, c_uint, c_void};
+use std::os::raw::{c_int, c_uint, c_void};
 use std::sync::OnceLock;
 pub type CUresult = c_uint; // CUDA_SUCCESS == 0
@@ -20,8 +20,6 @@ pub type CUdeviceptr = u64;
 pub type CUgraphicsResource = *mut c_void;
 pub type CUarray = *mut c_void;
 pub type CUexternalMemory = *mut c_void; // opaque CUextMemory_st*
 pub type CUmodule = *mut c_void; // opaque CUmod_st*
 pub type CUfunction = *mut c_void; // opaque CUfunc_st*
 /// `CUmemorytype` (cuda.h): HOST=1, DEVICE=2, ARRAY=3, UNIFIED=4.
 pub const CU_MEMORYTYPE_DEVICE: c_uint = 2;
@@ -144,26 +142,6 @@ pub(crate) struct CudaApi {
    cuIpcGetMemHandle: unsafe extern "C" fn(*mut CUipcMemHandle, CUdeviceptr) -> CUresult,
    cuIpcOpenMemHandle: unsafe extern "C" fn(*mut CUdeviceptr, CUipcMemHandle, c_uint) -> CUresult,
    cuIpcCloseMemHandle: unsafe extern "C" fn(CUdeviceptr) -> CUresult,
    // Cursor-overlay blend: a linear device alloc + a PTX module with the blend kernels launched
    // over the cursor's small rectangle (see [`CursorBlend`]).
    cuMemAlloc_v2: unsafe extern "C" fn(*mut CUdeviceptr, usize) -> CUresult,
    cuModuleLoadData: unsafe extern "C" fn(*mut CUmodule, *const c_void) -> CUresult,
    cuModuleUnload: unsafe extern "C" fn(CUmodule) -> CUresult,
    cuModuleGetFunction: unsafe extern "C" fn(*mut CUfunction, CUmodule, *const c_char) -> CUresult,
    #[allow(clippy::type_complexity)]
    cuLaunchKernel: unsafe extern "C" fn(
        CUfunction,
        c_uint,
        c_uint,
        c_uint,
        c_uint,
        c_uint,
        c_uint,
        c_uint,
        CUstream,
        *mut *mut c_void,
        *mut *mut c_void,
    ) -> CUresult,
 }
 // SAFETY: every field is a bare `extern "C" fn` address into the leaked, process-lifetime
 // `libcuda` mapping (`cuda_api` `forget`s the `Library`, so it is never unloaded) — an immutable
@@ -236,11 +214,6 @@ pub(crate) fn cuda_api() -> Option<&'static CudaApi> {
                    .or_else(|_| lib.get(b"cuIpcOpenMemHandle\0"))
                    .ok()?,
                cuIpcCloseMemHandle: *lib.get(b"cuIpcCloseMemHandle\0").ok()?,
                cuMemAlloc_v2: *lib.get(b"cuMemAlloc_v2\0").ok()?,
                cuModuleLoadData: *lib.get(b"cuModuleLoadData\0").ok()?,
                cuModuleUnload: *lib.get(b"cuModuleUnload\0").ok()?,
                cuModuleGetFunction: *lib.get(b"cuModuleGetFunction\0").ok()?,
                cuLaunchKernel: *lib.get(b"cuLaunchKernel\0").ok()?,
            };
            std::mem::forget(lib); // keep libcuda mapped for the fn pointers' lifetime (process)
            Some(api)
@@ -304,53 +277,6 @@ pub(crate) unsafe fn cuMemFree_v2(dptr: CUdeviceptr) -> CUresult {
        None => CU_ERROR_NOT_LOADED,
    }
 }
 pub(crate) unsafe fn cuMemAlloc_v2(dptr: *mut CUdeviceptr, size: usize) -> CUresult {
    match cuda_api() {
        Some(a) => (a.cuMemAlloc_v2)(dptr, size),
        None => CU_ERROR_NOT_LOADED,
    }
 }
 pub(crate) unsafe fn cuModuleLoadData(m: *mut CUmodule, image: *const c_void) -> CUresult {
    match cuda_api() {
        Some(a) => (a.cuModuleLoadData)(m, image),
        None => CU_ERROR_NOT_LOADED,
    }
 }
 pub(crate) unsafe fn cuModuleUnload(m: CUmodule) -> CUresult {
    match cuda_api() {
        Some(a) => (a.cuModuleUnload)(m),
        None => CU_ERROR_NOT_LOADED,
    }
 }
 pub(crate) unsafe fn cuModuleGetFunction(
    f: *mut CUfunction,
    m: CUmodule,
    name: *const c_char,
 ) -> CUresult {
    match cuda_api() {
        Some(a) => (a.cuModuleGetFunction)(f, m, name),
        None => CU_ERROR_NOT_LOADED,
    }
 }
 #[allow(clippy::too_many_arguments)]
 pub(crate) unsafe fn cuLaunchKernel(
    f: CUfunction,
    gx: c_uint,
    gy: c_uint,
    gz: c_uint,
    bx: c_uint,
    by: c_uint,
    bz: c_uint,
    shmem: c_uint,
    stream: CUstream,
    params: *mut *mut c_void,
    extra: *mut *mut c_void,
 ) -> CUresult {
    match cuda_api() {
        Some(a) => (a.cuLaunchKernel)(f, gx, gy, gz, bx, by, bz, shmem, stream, params, extra),
        None => CU_ERROR_NOT_LOADED,
    }
 }
 pub(crate) unsafe fn cuMemcpy2DAsync_v2(copy: *const CUDA_MEMCPY2D, stream: CUstream) -> CUresult {
    match cuda_api() {
        Some(a) => (a.cuMemcpy2DAsync_v2)(copy, stream),
@@ -0,0 +1,170 @@
 #version 450
 // Cursor-overlay blend for the direct-SDK NVENC path (cursor-as-metadata), dispatched over the
 // cursor's rectangle only — the Vulkan replacement for the retired cursor_blend.cu PTX kernels
 // (PTX is JIT'd against the driver's ISA ceiling, so a vendored blob silently dies on older
 // drivers: CUDA errors 222/218 on-glass; SPIR-V has no such coupling). The NVENC input surface is
 // Vulkan-allocated, CUDA-imported external memory (see vkslot.rs), so this shader writes the very
 // bytes NVENC encodes.
 //
 // MODE (spec constant): 0 = packed 4-byte ARGB (NVENC byte order B,G,R,A), 1 = NV12 (Y plane +
 // interleaved half-res UV at row surfH), 2 = planar YUV444 (3 full-res planes stacked at
 // pitch*surfH). BT.709 limited-range coefficients — identical to rgb2nv12_buf.comp and the
 // retired .cu, so the cursor colour matches the frame regardless of backend.
 //
 // The surface SSBO is uint[] (no 8-bit storage dependency — maximum driver reach): every
 // invocation exclusively owns the 32-bit words it read-modify-writes. ARGB: one invocation per
 // cursor pixel = one word. NV12/YUV444: one invocation per WORD-ALIGNED 4-px luma span (per two
 // rows for NV12, whose 2 chroma bytes-pairs land in one exclusive word). Spans are aligned to the
 // SURFACE, not the cursor, so neighbouring invocations never share a word even at odd `ox`.
 //
 // Rebuild: glslc cursor_blend.comp -o cursor_blend.spv   (vendored beside this file)
 layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
 layout(constant_id = 0) const uint MODE = 0;
 layout(std430, binding = 0) buffer Surf { uint surf[]; };
 layout(std430, binding = 1) readonly buffer Cur { uint cur[]; };
 layout(push_constant) uniform Push {
    uint pitch;   // surface row stride, bytes (4-aligned by construction)
    uint surfW;   // content width, px
    uint surfH;   // luma rows (plane stride multiplier)
    uint curW;    // cursor bitmap width, px
    uint curH;    // cursor bitmap height, px
    int  ox;      // cursor top-left on the surface, px (may be negative)
    int  oy;
 } pc;
 // Cursor texel (straight-alpha RGBA, tight rows) or (0,0,0,0) outside the bitmap.
 uvec4 cursor_px(int cx, int cy) {
    if (cx < 0 || cy < 0 || cx >= int(pc.curW) || cy >= int(pc.curH)) return uvec4(0);
    uint w = cur[uint(cy) * pc.curW + uint(cx)];
    return uvec4(w & 0xFFu, (w >> 8) & 0xFFu, (w >> 16) & 0xFFu, (w >> 24) & 0xFFu); // R,G,B,A
 }
 uint blend8(uint dst, uint src, uint a) {
    return (src * a + dst * (255u - a)) / 255u;
 }
 // BT.709 limited RGB→Y/U/V (matches the retired .cu / rgb2nv12_buf.comp).
 uint y_of(uvec4 s) {
    return uint(clamp(16.0 + 0.1826 * float(s.r) + 0.6142 * float(s.g) + 0.0620 * float(s.b) + 0.5, 0.0, 255.0));
 }
 float u_of(uvec4 s) { return 128.0 - 0.1006 * float(s.r) - 0.3386 * float(s.g) + 0.4392 * float(s.b); }
 float v_of(uvec4 s) { return 128.0 + 0.4392 * float(s.r) - 0.3989 * float(s.g) - 0.0403 * float(s.b); }
 // Read-modify-write one byte lane of a word index.
 void rmw_byte(uint word_idx, uint lane, uint val8, uint a) {
    uint w = surf[word_idx];
    uint shift = lane * 8u;
    uint d = (w >> shift) & 0xFFu;
    uint b = blend8(d, val8, a);
    surf[word_idx] = (w & ~(0xFFu << shift)) | (b << shift);
 }
 void main() {
    if (MODE == 0u) {
        // ARGB: one invocation per cursor pixel; each surface pixel is one exclusive word.
        int cx = int(gl_GlobalInvocationID.x);
        int cy = int(gl_GlobalInvocationID.y);
        if (cx >= int(pc.curW) || cy >= int(pc.curH)) return;
        int px = pc.ox + cx, py = pc.oy + cy;
        if (px < 0 || py < 0 || px >= int(pc.surfW) || py >= int(pc.surfH)) return;
        uvec4 s = cursor_px(cx, cy);
        if (s.a == 0u) return;
        uint idx = (uint(py) * pc.pitch + uint(px) * 4u) / 4u;
        uint w = surf[idx];
        uint b = blend8(w & 0xFFu, s.b, s.a);           // B lane
        uint g = blend8((w >> 8) & 0xFFu, s.g, s.a);    // G lane
        uint r = blend8((w >> 16) & 0xFFu, s.r, s.a);   // R lane
        surf[idx] = (w & 0xFF000000u) | (r << 16) | (g << 8) | b;
        return;
    }
    // NV12 / YUV444: one invocation per SURFACE-word-aligned 4-px luma span. Span origin:
    // x0 = floor(ox/4)*4 + span*4 (surface px), rows walk the cursor rect.
    int span = int(gl_GlobalInvocationID.x);
    int row = int(gl_GlobalInvocationID.y);
    int x0 = (pc.ox >> 2) << 2;         // word-aligned start at/left-of ox (ox may be negative)
    int px0 = x0 + span * 4;
    if (MODE == 2u) {
        // YUV444: rows walk cursor rows one at a time.
        if (row >= int(pc.curH)) return;
        int py = pc.oy + row;
        if (py < 0 || py >= int(pc.surfH)) return;
        uint plane = pc.pitch * pc.surfH;
        for (int i = 0; i < 4; i++) {
            int px = px0 + i;
            int cx = px - pc.ox;
            if (px < 0 || px >= int(pc.surfW)) continue;
            uvec4 s = cursor_px(cx, row);
            if (s.a == 0u) continue;
            uint off = uint(py) * pc.pitch + uint(px);
            uint U = uint(clamp(u_of(s) + 0.5, 0.0, 255.0));
            uint V = uint(clamp(v_of(s) + 0.5, 0.0, 255.0));
            rmw_byte(off / 4u, off % 4u, y_of(s), s.a);
            rmw_byte((plane + off) / 4u, (plane + off) % 4u, U, s.a);
            rmw_byte((2u * plane + off) / 4u, (2u * plane + off) % 4u, V, s.a);
        }
        return;
    }
    // NV12: rows walk 2-row luma blocks (row = block row). The span's 4 luma px × 2 rows are
    // exclusive words; its 2 chroma samples (4 bytes) are one exclusive word.
    int base_cy = row * 2;
    if (base_cy >= int(pc.curH)) return;
    // Luma: 4 px × 2 rows.
    for (int j = 0; j < 2; j++) {
        int cy = base_cy + j;
        int py = pc.oy + cy;
        if (cy >= int(pc.curH) || py < 0 || py >= int(pc.surfH)) continue;
        for (int i = 0; i < 4; i++) {
            int px = px0 + i;
            int cx = px - pc.ox;
            if (px < 0 || px >= int(pc.surfW)) continue;
            uvec4 s = cursor_px(cx, cy);
            if (s.a == 0u) continue;
            uint off = uint(py) * pc.pitch + uint(px);
            rmw_byte(off / 4u, off % 4u, y_of(s), s.a);
        }
    }
    // Chroma: two UV samples covering the span's 2x2 blocks, alpha-weighted like the .cu kernel.
    // The UV plane starts at row surfH; sample (uvx, uvy) lives at uv_base + uvy*pitch + uvx*2.
    // Guard: only spans whose px0 is 4-aligned own their chroma word (px0 is by construction).
    int py_top = pc.oy + base_cy;
    int uvy = py_top >> 1;
    if (py_top < 0 || uvy < 0 || uvy * 2 >= int(pc.surfH)) return;
    uint uv_base = pc.pitch * pc.surfH;
    for (int hf = 0; hf < 2; hf++) {
        // Each hf = one 2x2 luma block = one UV sample (2 bytes).
        int bx = px0 + hf * 2;
        if (bx < 0 || bx >= int(pc.surfW)) continue;
        int uvx = bx >> 1;
        float ua = 0.0, va = 0.0, wa = 0.0;
        int cnt = 0;
        for (int j = 0; j < 2; j++) {
            for (int i = 0; i < 2; i++) {
                int px = bx + i;
                int py = py_top + j;
                int cx = px - pc.ox;
                int cy = base_cy + j;
                if (px < 0 || py < 0 || px >= int(pc.surfW) || py >= int(pc.surfH)) continue;
                uvec4 s = cursor_px(cx, cy);
                if (s.a == 0u) continue;
                ua += u_of(s) * float(s.a);
                va += v_of(s) * float(s.a);
                wa += float(s.a);
                cnt++;
            }
        }
        if (wa <= 0.0 || cnt == 0) continue;
        uint U = uint(clamp(ua / wa + 0.5, 0.0, 255.0));
        uint V = uint(clamp(va / wa + 0.5, 0.0, 255.0));
        uint amean = uint(clamp(wa / float(cnt) + 0.5, 0.0, 255.0));
        uint off = uv_base + uint(uvy) * pc.pitch + uint(uvx) * 2u;
        rmw_byte(off / 4u, off % 4u, U, amean);
        rmw_byte((off + 1u) / 4u, (off + 1u) % 4u, V, amean);
    }
 }
@@ -15,6 +15,7 @@ pub mod client;
 pub mod cuda;
 pub mod egl;
 pub mod proto;
 pub mod vkslot;
 pub mod vulkan;
 pub mod worker;
@@ -0,0 +1,710 @@
 //! Vulkan-allocated NVENC input slots + the Vulkan compute cursor blend — the driver-portable
 //! replacement for the retired `cursor_blend.cu` PTX kernels (design: remote-desktop-sweep §8,
 //! Phase A). A vendored PTX blob is JIT'd against the driver's ISA ceiling, so it silently dies
 //! on drivers older than the generating toolkit (CUDA errors 222/218 on-glass — the KWin leg's
 //! invisible composite cursor); SPIR-V has no such coupling.
 //!
 //! ```text
 //!   exportable VkBuffer ──vkGetMemoryFdKHR(OPAQUE_FD)──▶ cuImportExternalMemory ──▶ CUdeviceptr
 //!        ▲                                                     │ NVENC registers + encodes
 //!        └── cursor_blend.comp dispatch (cursor rect only) ◀───┘ CUDA copies frames in
 //! ```
 //!
 //! The direct-SDK NVENC encoder allocates its input ring through [`VkSlotBlend::alloc_slot`]
 //! instead of `cuMemAllocPitch`: same contiguous layouts (`InputSurface` docs), but the memory is
 //! Vulkan external memory both APIs address. Per cursor-bearing frame the encoder CPU-syncs its
 //! CUDA copy, then [`VkSlotBlend::blend`] dispatches the compute blend over the cursor's
 //! rectangle and fence-waits — the same coherence ceremony [`super::vulkan::VkBridge`] ships for
 //! its CSC (fence-ordered cross-API access on NVIDIA, no queue-family transfer needed). Frames
 //! without a cursor never touch Vulkan, keeping the stream-ordered fast path intact.
 //!
 //! Falls back cleanly: if bring-up fails the encoder allocates plain CUDA surfaces and composite
 //! mode degrades to no cursor (warned once) — never a failed session.
 // Every `unsafe` block in this file carries a `// SAFETY:` proof; enforce it (unsafe-proof program).
 #![deny(clippy::undocumented_unsafe_blocks)]
 use super::cuda::{self, CUdeviceptr};
 use anyhow::{anyhow, Context as _, Result};
 use ash::vk;
 /// Max cursor-overlay bitmap edge (px) — matches [`cuda::CURSOR_MAX`] and the capture-side clamp.
 pub const CURSOR_MAX: u32 = cuda::CURSOR_MAX;
 /// The vendored SPIR-V for `cursor_blend.comp` (beside this file; rebuild with
 /// `glslc cursor_blend.comp -o cursor_blend.spv`).
 const CURSOR_SPV: &[u8] = include_bytes!("cursor_blend.spv");
 /// NVENC input-surface layout — selects the spec-constant `MODE` pipeline and the allocation
 /// arithmetic (mirroring `InputSurface`'s contiguous layouts).
 #[derive(Clone, Copy, PartialEq, Eq, Debug)]
 pub enum SlotFormat {
    /// Packed 4-byte ARGB (NVENC byte order B,G,R,A): `pitch × height`.
    Argb,
    /// NV12: Y rows `[0, H)` + interleaved UV rows `[H, 3H/2)` under one pitch.
    Nv12,
    /// Planar YUV444: three full-res planes stacked at `pitch × height` intervals.
    Yuv444,
 }
 impl SlotFormat {
    fn mode(self) -> u32 {
        match self {
            SlotFormat::Argb => 0,
            SlotFormat::Nv12 => 1,
            SlotFormat::Yuv444 => 2,
        }
    }
    fn row_bytes(self, width: u32) -> u64 {
        match self {
            SlotFormat::Argb => width as u64 * 4,
            SlotFormat::Nv12 | SlotFormat::Yuv444 => width as u64,
        }
    }
    fn rows(self, height: u32) -> u64 {
        match self {
            SlotFormat::Argb => height as u64,
            SlotFormat::Nv12 => height as u64 + (height as u64 / 2).max(1),
            SlotFormat::Yuv444 => height as u64 * 3,
        }
    }
 }
 /// What the encoder holds per ring slot: the CUDA view it registers with NVENC plus the id it
 /// hands back to [`VkSlotBlend::blend`]. The backing Vulkan objects + CUDA mapping live in the
 /// [`VkSlotBlend`] (freed by [`free_slots`](VkSlotBlend::free_slots) / drop), so this is Copy —
 /// the encoder's ring keeps its existing shape.
 #[derive(Clone, Copy)]
 pub struct VkSlotRef {
    /// Device pointer NVENC registers (CUDA's mapping of the Vulkan memory).
    pub ptr: CUdeviceptr,
    /// Row stride in bytes (ours: row bytes rounded up to 256).
    pub pitch: usize,
    /// Luma rows (the plane-stride multiplier, as in `InputSurface`).
    pub height: u32,
    /// Index into the blend's slot table.
    pub id: usize,
 }
 /// One allocated slot's backing objects, freed together in reverse order (CUDA mapping first).
 struct SlotAlloc {
    buffer: vk::Buffer,
    memory: vk::DeviceMemory,
    /// CUDA's import of the exported OPAQUE_FD — must drop BEFORE the Vulkan memory is freed.
    cuda: cuda::ExternalDmabuf,
    size: u64,
 }
 /// 28-byte push-constant block matching `cursor_blend.comp`'s `Push`.
 #[repr(C)]
 struct Push {
    pitch: u32,
    surf_w: u32,
    surf_h: u32,
    cur_w: u32,
    cur_h: u32,
    ox: i32,
    oy: i32,
 }
 pub struct VkSlotBlend {
    _entry: ash::Entry,
    instance: ash::Instance,
    device: ash::Device,
    ext_fd: ash::khr::external_memory_fd::Device,
    queue: vk::Queue,
    cmd_pool: vk::CommandPool,
    cmd: vk::CommandBuffer,
    fence: vk::Fence,
    mem_props: vk::PhysicalDeviceMemoryProperties,
    shader: vk::ShaderModule,
    desc_layout: vk::DescriptorSetLayout,
    pipe_layout: vk::PipelineLayout,
    desc_pool: vk::DescriptorPool,
    desc_set: vk::DescriptorSet,
    /// One pipeline per [`SlotFormat`], indexed by `mode()` (spec constant).
    pipelines: [vk::Pipeline; 3],
    /// Host-visible cursor bitmap staging (CURSOR_MAX²·4, tight rows), persistently mapped.
    cur_buf: vk::Buffer,
    cur_mem: vk::DeviceMemory,
    cur_map: *mut u8,
    slots: Vec<SlotAlloc>,
 }
 // SAFETY: raw Vulkan handles + a persistently-mapped pointer, all uniquely owned by this struct
 // and destroyed exactly once in `Drop`; used from the encoder thread but moved with it. `Send`
 // only (not `Sync`), matching the single-thread use — transferring opaque handles cannot dangle.
 unsafe impl Send for VkSlotBlend {}
 impl VkSlotBlend {
    /// Bring up the device + blend pipelines. Requires the CUDA shared context (the encoder's) to
    /// be established; picks the NVIDIA physical device (the NVENC path is NVIDIA by definition).
    pub fn new() -> Result<VkSlotBlend> {
        // SAFETY: standard ash bring-up, same shape as `VkBridge::new` — every call is `unsafe`
        // only because ash cannot statically verify handle/CreateInfo validity. Every
        // `*CreateInfo`/`AllocateInfo` is built from locals that live for the duration of the
        // synchronous call reading them; every handle passed was created and `?`-checked in this
        // same function. Shares nothing across threads.
        unsafe {
            let entry = ash::Entry::load().context("load libvulkan")?;
            let app = vk::ApplicationInfo::default().api_version(vk::API_VERSION_1_1);
            let instance = entry
                .create_instance(
                    &vk::InstanceCreateInfo::default().application_info(&app),
                    None,
                )
                .context("vkCreateInstance")?;
            let phys = match instance
                .enumerate_physical_devices()
                .context("enumerate GPUs")?
                .into_iter()
                .find(|&p| instance.get_physical_device_properties(p).vendor_id == 0x10DE)
            {
                Some(p) => p,
                None => {
                    instance.destroy_instance(None);
                    return Err(anyhow!("no NVIDIA Vulkan device"));
                }
            };
            let mem_props = instance.get_physical_device_memory_properties(phys);
            let qf = match instance
                .get_physical_device_queue_family_properties(phys)
                .iter()
                .position(|q| q.queue_flags.contains(vk::QueueFlags::COMPUTE))
            {
                Some(i) => i as u32,
                None => {
                    instance.destroy_instance(None);
                    return Err(anyhow!("no compute-capable queue family"));
                }
            };
            let prio = [1.0f32];
            let qci = [vk::DeviceQueueCreateInfo::default()
                .queue_family_index(qf)
                .queue_priorities(&prio)];
            let exts = [ash::khr::external_memory_fd::NAME.as_ptr()];
            let device = match instance.create_device(
                phys,
                &vk::DeviceCreateInfo::default()
                    .queue_create_infos(&qci)
                    .enabled_extension_names(&exts),
                None,
            ) {
                Ok(d) => d,
                Err(e) => {
                    instance.destroy_instance(None);
                    return Err(e).context("vkCreateDevice (external_memory_fd supported?)");
                }
            };
            // From here teardown-on-error goes through `destroy_partial`, which tolerates null
            // handles — build everything into an incrementally-filled struct.
            let ext_fd = ash::khr::external_memory_fd::Device::new(&instance, &device);
            let queue = device.get_device_queue(qf, 0);
            let mut me = VkSlotBlend {
                _entry: entry,
                instance,
                device,
                ext_fd,
                queue,
                cmd_pool: vk::CommandPool::null(),
                cmd: vk::CommandBuffer::null(),
                fence: vk::Fence::null(),
                mem_props,
                shader: vk::ShaderModule::null(),
                desc_layout: vk::DescriptorSetLayout::null(),
                pipe_layout: vk::PipelineLayout::null(),
                desc_pool: vk::DescriptorPool::null(),
                desc_set: vk::DescriptorSet::null(),
                pipelines: [vk::Pipeline::null(); 3],
                cur_buf: vk::Buffer::null(),
                cur_mem: vk::DeviceMemory::null(),
                cur_map: std::ptr::null_mut(),
                slots: Vec::new(),
            };
            me.init_objects(qf).inspect_err(|_| {
                // `Drop` runs the same teardown and tolerates the nulls left by a partial init.
            })?;
            tracing::info!(
                "Vulkan slot blend ready (exportable NVENC inputs + SPIR-V cursor blend)"
            );
            Ok(me)
        }
    }
    /// The non-device objects: command machinery, cursor staging, descriptor + pipelines.
    fn init_objects(&mut self, qf: u32) -> Result<()> {
        // SAFETY: same contract as `new` — ash calls on the live `self.device` with builder infos
        // from locals outliving each synchronous call; created handles are stored into `self`
        // immediately so the caller's `Drop` frees them on any later failure.
        unsafe {
            let d = &self.device;
            self.cmd_pool = d
                .create_command_pool(
                    &vk::CommandPoolCreateInfo::default()
                        .queue_family_index(qf)
                        .flags(vk::CommandPoolCreateFlags::RESET_COMMAND_BUFFER),
                    None,
                )
                .context("create command pool")?;
            self.cmd = d
                .allocate_command_buffers(
                    &vk::CommandBufferAllocateInfo::default()
                        .command_pool(self.cmd_pool)
                        .level(vk::CommandBufferLevel::PRIMARY)
                        .command_buffer_count(1),
                )
                .context("allocate command buffer")?[0];
            self.fence = d
                .create_fence(&vk::FenceCreateInfo::default(), None)
                .context("create fence")?;
            // Cursor staging: host-visible+coherent SSBO, persistently mapped.
            let cur_size = (CURSOR_MAX * CURSOR_MAX * 4) as u64;
            self.cur_buf = d
                .create_buffer(
                    &vk::BufferCreateInfo::default()
                        .size(cur_size)
                        .usage(vk::BufferUsageFlags::STORAGE_BUFFER),
                    None,
                )
                .context("create cursor buffer")?;
            let reqs = d.get_buffer_memory_requirements(self.cur_buf);
            let mem_type = self
                .memory_type(
                    reqs.memory_type_bits,
                    vk::MemoryPropertyFlags::HOST_VISIBLE | vk::MemoryPropertyFlags::HOST_COHERENT,
                )
                .context("cursor buffer memory type")?;
            self.cur_mem = d
                .allocate_memory(
                    &vk::MemoryAllocateInfo::default()
                        .allocation_size(reqs.size)
                        .memory_type_index(mem_type),
                    None,
                )
                .context("allocate cursor memory")?;
            d.bind_buffer_memory(self.cur_buf, self.cur_mem, 0)
                .context("bind cursor memory")?;
            self.cur_map = d
                .map_memory(self.cur_mem, 0, cur_size, vk::MemoryMapFlags::empty())
                .context("map cursor memory")? as *mut u8;
            // Descriptor set: binding 0 = surface SSBO (rebound per blend), 1 = cursor SSBO.
            let bindings = [
                vk::DescriptorSetLayoutBinding::default()
                    .binding(0)
                    .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
                    .descriptor_count(1)
                    .stage_flags(vk::ShaderStageFlags::COMPUTE),
                vk::DescriptorSetLayoutBinding::default()
                    .binding(1)
                    .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
                    .descriptor_count(1)
                    .stage_flags(vk::ShaderStageFlags::COMPUTE),
            ];
            self.desc_layout = d
                .create_descriptor_set_layout(
                    &vk::DescriptorSetLayoutCreateInfo::default().bindings(&bindings),
                    None,
                )
                .context("create descriptor layout")?;
            let pc = [vk::PushConstantRange::default()
                .stage_flags(vk::ShaderStageFlags::COMPUTE)
                .size(std::mem::size_of::<Push>() as u32)];
            let dl = [self.desc_layout];
            self.pipe_layout = d
                .create_pipeline_layout(
                    &vk::PipelineLayoutCreateInfo::default()
                        .set_layouts(&dl)
                        .push_constant_ranges(&pc),
                    None,
                )
                .context("create pipeline layout")?;
            let pool_sizes = [vk::DescriptorPoolSize::default()
                .ty(vk::DescriptorType::STORAGE_BUFFER)
                .descriptor_count(2)];
            self.desc_pool = d
                .create_descriptor_pool(
                    &vk::DescriptorPoolCreateInfo::default()
                        .max_sets(1)
                        .pool_sizes(&pool_sizes),
                    None,
                )
                .context("create descriptor pool")?;
            let dls = [self.desc_layout];
            self.desc_set = d
                .allocate_descriptor_sets(
                    &vk::DescriptorSetAllocateInfo::default()
                        .descriptor_pool(self.desc_pool)
                        .set_layouts(&dls),
                )
                .context("allocate descriptor set")?[0];
            // The shader + one pipeline per MODE (spec constant 0).
            if CURSOR_SPV.len() % 4 != 0 {
                anyhow::bail!("cursor_blend.spv is not word-aligned");
            }
            let words: Vec<u32> = CURSOR_SPV
                .chunks_exact(4)
                .map(|c| u32::from_le_bytes([c[0], c[1], c[2], c[3]]))
                .collect();
            self.shader = d
                .create_shader_module(&vk::ShaderModuleCreateInfo::default().code(&words), None)
                .context("create blend shader module")?;
            for mode in 0u32..3 {
                let entries = [vk::SpecializationMapEntry::default()
                    .constant_id(0)
                    .offset(0)
                    .size(4)];
                let data = mode.to_le_bytes();
                let spec = vk::SpecializationInfo::default()
                    .map_entries(&entries)
                    .data(&data);
                let stage = vk::PipelineShaderStageCreateInfo::default()
                    .stage(vk::ShaderStageFlags::COMPUTE)
                    .module(self.shader)
                    .name(c"main")
                    .specialization_info(&spec);
                let info = [vk::ComputePipelineCreateInfo::default()
                    .stage(stage)
                    .layout(self.pipe_layout)];
                let p = d
                    .create_compute_pipelines(vk::PipelineCache::null(), &info, None)
                    .map_err(|(_, e)| e)
                    .context("create blend pipeline")?[0];
                self.pipelines[mode as usize] = p;
            }
        }
        Ok(())
    }
    fn memory_type(&self, type_bits: u32, flags: vk::MemoryPropertyFlags) -> Result<u32> {
        (0..self.mem_props.memory_type_count)
            .find(|&i| {
                type_bits & (1 << i) != 0
                    && self.mem_props.memory_types[i as usize]
                        .property_flags
                        .contains(flags)
            })
            .ok_or_else(|| anyhow!("no memory type for flags {flags:?}"))
    }
    /// Allocate one NVENC input slot as exportable Vulkan memory mapped into CUDA. Layout matches
    /// `InputSurface` (contiguous planes under one pitch); pitch = row bytes rounded to 256.
    pub fn alloc_slot(&mut self, fmt: SlotFormat, width: u32, height: u32) -> Result<VkSlotRef> {
        let pitch = (fmt.row_bytes(width) + 255) & !255;
        let size = pitch * fmt.rows(height);
        // SAFETY: exportable-buffer allocation, the exact `VkBridge::ensure_dst` incantation:
        // `ExternalMemoryBufferCreateInfo`/`ExportMemoryAllocateInfo` declare OPAQUE_FD,
        // `MemoryDedicatedAllocateInfo` ties the memory to the buffer; every info is a local
        // outliving its synchronous call and every failure path destroys the objects created so
        // far exactly once. `get_memory_fd` hands us an fd that `import_owned_fd` either adopts
        // (driver owns it) or closes on failure.
        unsafe {
            let d = &self.device;
            let mut ext_info = vk::ExternalMemoryBufferCreateInfo::default()
                .handle_types(vk::ExternalMemoryHandleTypeFlags::OPAQUE_FD);
            let buffer = d
                .create_buffer(
                    &vk::BufferCreateInfo::default()
                        .size(size)
                        .usage(vk::BufferUsageFlags::STORAGE_BUFFER)
                        .push_next(&mut ext_info),
                    None,
                )
                .context("create slot buffer")?;
            let reqs = d.get_buffer_memory_requirements(buffer);
            let mem_type = match self
                .memory_type(reqs.memory_type_bits, vk::MemoryPropertyFlags::DEVICE_LOCAL)
            {
                Ok(t) => t,
                Err(e) => {
                    d.destroy_buffer(buffer, None);
                    return Err(e);
                }
            };
            let mut export = vk::ExportMemoryAllocateInfo::default()
                .handle_types(vk::ExternalMemoryHandleTypeFlags::OPAQUE_FD);
            let mut dedicated = vk::MemoryDedicatedAllocateInfo::default().buffer(buffer);
            let memory = match d.allocate_memory(
                &vk::MemoryAllocateInfo::default()
                    .allocation_size(reqs.size)
                    .memory_type_index(mem_type)
                    .push_next(&mut export)
                    .push_next(&mut dedicated),
                None,
            ) {
                Ok(m) => m,
                Err(e) => {
                    d.destroy_buffer(buffer, None);
                    return Err(e).context("allocate exportable slot memory");
                }
            };
            if let Err(e) = d.bind_buffer_memory(buffer, memory, 0) {
                d.free_memory(memory, None);
                d.destroy_buffer(buffer, None);
                return Err(e).context("bind slot memory");
            }
            let fd = match self.ext_fd.get_memory_fd(
                &vk::MemoryGetFdInfoKHR::default()
                    .memory(memory)
                    .handle_type(vk::ExternalMemoryHandleTypeFlags::OPAQUE_FD),
            ) {
                Ok(f) => f,
                Err(e) => {
                    d.free_memory(memory, None);
                    d.destroy_buffer(buffer, None);
                    return Err(e).context("vkGetMemoryFdKHR(slot)");
                }
            };
            let ext = match cuda::ExternalDmabuf::import_owned_fd(fd, reqs.size) {
                Ok(c) => c,
                Err(e) => {
                    d.free_memory(memory, None);
                    d.destroy_buffer(buffer, None);
                    return Err(e).context("cuImportExternalMemory(slot OPAQUE_FD)");
                }
            };
            let r = VkSlotRef {
                ptr: ext.ptr,
                pitch: pitch as usize,
                height,
                id: self.slots.len(),
            };
            self.slots.push(SlotAlloc {
                buffer,
                memory,
                cuda: ext,
                size,
            });
            Ok(r)
        }
    }
    /// Free every allocated slot (encoder teardown, alongside its ring clear). CUDA mappings drop
    /// first (field order in [`SlotAlloc`] frees `cuda` via its own `Drop` before we free the VK
    /// objects explicitly here).
    pub fn free_slots(&mut self) {
        for s in self.slots.drain(..) {
            drop(s.cuda); // CUDA's view of the memory goes first
                          // SAFETY: `buffer`/`memory` were created in `alloc_slot`, are uniquely owned by the
                          // drained `SlotAlloc`, and are destroyed exactly once here. No queue work can be
                          // in flight: every `blend` fence-waits before returning.
            unsafe {
                self.device.destroy_buffer(s.buffer, None);
                self.device.free_memory(s.memory, None);
            }
        }
    }
    /// Upload the cursor RGBA (`cw*ch*4`, tight rows) into the mapped staging buffer. Call only
    /// when the bitmap changes; position moves are push constants.
    pub fn upload_cursor(&mut self, rgba: &[u8], cw: u32, ch: u32) {
        let cw = cw.min(CURSOR_MAX);
        let ch = ch.min(CURSOR_MAX);
        let len = (cw * ch * 4) as usize;
        let len = len.min(rgba.len());
        // SAFETY: `cur_map` is the live persistent mapping of the CURSOR_MAX²·4 host-coherent
        // allocation (created in `init_objects`, unmapped only in `Drop`); `len` is clamped to
        // both the source slice and the buffer capacity. No blend is in flight (every `blend`
        // fence-waits before returning), so no GPU read races this host write.
        unsafe {
            std::ptr::copy_nonoverlapping(rgba.as_ptr(), self.cur_map, len);
        }
    }
    /// Blend the uploaded cursor into `slot` at `(ox, oy)`: record, submit, fence-wait. The
    /// caller has CPU-synced its CUDA frame copy first; the fence wait makes the shader's writes
    /// visible to the subsequent NVENC encode (the `VkBridge` precedent: fence-ordered cross-API
    /// access, no queue-family transfer — NVIDIA-only path).
    #[allow(clippy::too_many_arguments)] // surface geometry + cursor rect — unpacked kernel args
    pub fn blend_ref(
        &mut self,
        slot: &VkSlotRef,
        fmt: SlotFormat,
        surf_w: u32,
        cw: u32,
        ch: u32,
        ox: i32,
        oy: i32,
    ) -> Result<()> {
        let alloc = self
            .slots
            .get(slot.id)
            .ok_or_else(|| anyhow!("bad slot id {}", slot.id))?;
        let cw = cw.min(CURSOR_MAX);
        let ch = ch.min(CURSOR_MAX);
        if cw == 0 || ch == 0 {
            return Ok(());
        }
        let push = Push {
            pitch: slot.pitch as u32,
            surf_w,
            surf_h: slot.height,
            cur_w: cw,
            cur_h: ch,
            ox,
            oy,
        };
        // Dispatch geometry (must match cursor_blend.comp): ARGB = per cursor px; NV12/YUV444 =
        // per word-aligned 4-px span × (2-row blocks | rows).
        let (gx, gy) = match fmt {
            SlotFormat::Argb => (cw.div_ceil(8), ch.div_ceil(8)),
            _ => {
                let x0 = (ox >> 2) << 2;
                let spans = ((ox + cw as i32) - x0 + 3).div_euclid(4).max(1) as u32;
                let rows = match fmt {
                    SlotFormat::Nv12 => ch.div_ceil(2),
                    _ => ch,
                };
                (spans.div_ceil(8), rows.div_ceil(8))
            }
        };
        // SAFETY: single-threaded record/submit/wait on handles this struct owns. The descriptor
        // update is safe because no prior submission is in flight (every blend fence-waits and
        // the fence is reset before reuse). Buffer infos and barrier structs are locals outliving
        // their synchronous calls. The dispatch's shader accesses stay in-bounds by the shader's
        // own guards (surfW/surfH/curW/curH from `push`) against the slot allocation sized in
        // `alloc_slot` for exactly that geometry.
        unsafe {
            let d = &self.device;
            let surf_info = [vk::DescriptorBufferInfo::default()
                .buffer(alloc.buffer)
                .offset(0)
                .range(alloc.size)];
            let cur_info = [vk::DescriptorBufferInfo::default()
                .buffer(self.cur_buf)
                .offset(0)
                .range((CURSOR_MAX * CURSOR_MAX * 4) as u64)];
            let writes = [
                vk::WriteDescriptorSet::default()
                    .dst_set(self.desc_set)
                    .dst_binding(0)
                    .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
                    .buffer_info(&surf_info),
                vk::WriteDescriptorSet::default()
                    .dst_set(self.desc_set)
                    .dst_binding(1)
                    .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
                    .buffer_info(&cur_info),
            ];
            d.update_descriptor_sets(&writes, &[]);
            d.begin_command_buffer(
                self.cmd,
                &vk::CommandBufferBeginInfo::default()
                    .flags(vk::CommandBufferUsageFlags::ONE_TIME_SUBMIT),
            )
            .context("begin blend cmd")?;
            // CUDA wrote the frame into this memory outside Vulkan's view — make it visible to
            // the shader (external-memory coherence ceremony; NVIDIA honors this with the fence
            // ordering alone, the barrier is the spec-shaped belt-and-braces).
            let acquire = [vk::MemoryBarrier::default()
                .src_access_mask(vk::AccessFlags::MEMORY_WRITE)
                .dst_access_mask(vk::AccessFlags::SHADER_READ | vk::AccessFlags::SHADER_WRITE)];
            d.cmd_pipeline_barrier(
                self.cmd,
                vk::PipelineStageFlags::TOP_OF_PIPE,
                vk::PipelineStageFlags::COMPUTE_SHADER,
                vk::DependencyFlags::empty(),
                &acquire,
                &[],
                &[],
            );
            d.cmd_bind_pipeline(
                self.cmd,
                vk::PipelineBindPoint::COMPUTE,
                self.pipelines[fmt.mode() as usize],
            );
            d.cmd_bind_descriptor_sets(
                self.cmd,
                vk::PipelineBindPoint::COMPUTE,
                self.pipe_layout,
                0,
                &[self.desc_set],
                &[],
            );
            let bytes = std::slice::from_raw_parts(
                (&push as *const Push) as *const u8,
                std::mem::size_of::<Push>(),
            );
            d.cmd_push_constants(
                self.cmd,
                self.pipe_layout,
                vk::ShaderStageFlags::COMPUTE,
                0,
                bytes,
            );
            d.cmd_dispatch(self.cmd, gx.max(1), gy.max(1), 1);
            // Release the shader's writes so the post-fence CUDA/NVENC reads see them.
            let release = [vk::MemoryBarrier::default()
                .src_access_mask(vk::AccessFlags::SHADER_WRITE)
                .dst_access_mask(vk::AccessFlags::MEMORY_READ)];
            d.cmd_pipeline_barrier(
                self.cmd,
                vk::PipelineStageFlags::COMPUTE_SHADER,
                vk::PipelineStageFlags::BOTTOM_OF_PIPE,
                vk::DependencyFlags::empty(),
                &release,
                &[],
                &[],
            );
            d.end_command_buffer(self.cmd).context("end blend cmd")?;
            let cmds = [self.cmd];
            let submit = [vk::SubmitInfo::default().command_buffers(&cmds)];
            d.queue_submit(self.queue, &submit, self.fence)
                .context("submit blend")?;
            let r = d.wait_for_fences(&[self.fence], true, 1_000_000_000);
            d.reset_fences(&[self.fence]).ok();
            r.context("blend fence wait")?;
        }
        Ok(())
    }
 }
 impl Drop for VkSlotBlend {
    fn drop(&mut self) {
        self.free_slots();
        // SAFETY: every handle below was created in `new`/`init_objects` (or is null from a
        // partial init — Vulkan destroy/free calls are defined no-ops on null handles) and is
        // uniquely owned; each is destroyed exactly once here, pipelines/layouts/pools before the
        // device, the device before the instance. No work is in flight (`blend` fence-waits).
        unsafe {
            let d = &self.device;
            for p in self.pipelines {
                if p != vk::Pipeline::null() {
                    d.destroy_pipeline(p, None);
                }
            }
            if self.shader != vk::ShaderModule::null() {
                d.destroy_shader_module(self.shader, None);
            }
            if self.desc_pool != vk::DescriptorPool::null() {
                d.destroy_descriptor_pool(self.desc_pool, None);
            }
            if self.pipe_layout != vk::PipelineLayout::null() {
                d.destroy_pipeline_layout(self.pipe_layout, None);
            }
            if self.desc_layout != vk::DescriptorSetLayout::null() {
                d.destroy_descriptor_set_layout(self.desc_layout, None);
            }
            if !self.cur_map.is_null() {
                d.unmap_memory(self.cur_mem);
            }
            if self.cur_buf != vk::Buffer::null() {
                d.destroy_buffer(self.cur_buf, None);
            }
            if self.cur_mem != vk::DeviceMemory::null() {
                d.free_memory(self.cur_mem, None);
            }
            if self.fence != vk::Fence::null() {
                d.destroy_fence(self.fence, None);
            }
            if self.cmd_pool != vk::CommandPool::null() {
                d.destroy_command_pool(self.cmd_pool, None);
            }
            d.destroy_device(None);
            self.instance.destroy_instance(None);
        }
    }
 }
@@ -555,6 +555,9 @@ pub struct PunktfunkConnection {
    /// (a fetched payload, an offer's format list, or a fetch-request's MIME) —
    /// borrow-until-next-call, same contract as `last`.
    last_clip: std::sync::Mutex<Option<Vec<u8>>>,
    /// The last cursor shape handed out — `next_cursor_shape`'s `rgba` pointer borrows it
    /// until the next cursor-shape call (the `last_audio` contract).
    last_cursor_shape: std::sync::Mutex<Option<crate::quic::CursorShape>>,
 }
 /// Lazily-initialized in-core Opus decode state. A coupled-1-stream multistream decoder is
@@ -867,6 +870,94 @@ impl PunktfunkRichInputEx {
    }
 }
 /// [`PunktfunkPenSample::state`] bit: the pen hovers in range (implied by `TOUCHING`).
 pub const PUNKTFUNK_PEN_IN_RANGE: u8 = 0x01;
 /// [`PunktfunkPenSample::state`] bit: the tip is in contact.
 pub const PUNKTFUNK_PEN_TOUCHING: u8 = 0x02;
 /// [`PunktfunkPenSample::state`] bit: primary barrel button (or squeeze mapping) held.
 pub const PUNKTFUNK_PEN_BARREL1: u8 = 0x04;
 /// [`PunktfunkPenSample::state`] bit: secondary barrel button (or double-tap mapping) held.
 pub const PUNKTFUNK_PEN_BARREL2: u8 = 0x08;
 /// [`PunktfunkPenSample::tool`]: the pen tip.
 pub const PUNKTFUNK_PEN_TOOL_PEN: u8 = 0;
 /// [`PunktfunkPenSample::tool`]: the eraser (a client-side mode — Apple Pencil has no
 /// hardware eraser end; the squeeze/double-tap mapping usually drives this).
 pub const PUNKTFUNK_PEN_TOOL_ERASER: u8 = 1;
 /// Most samples one [`punktfunk_connection_send_pen`] call accepts (one wire batch).
 pub const PUNKTFUNK_PEN_BATCH_MAX: u32 = 8;
 /// [`PunktfunkPenSample::tilt_deg`] sentinel: no tilt reading.
 pub const PUNKTFUNK_PEN_TILT_UNKNOWN: u8 = 0xFF;
 /// [`PunktfunkPenSample::azimuth_deg`] / `roll_deg` sentinel: no reading.
 pub const PUNKTFUNK_PEN_ANGLE_UNKNOWN: u16 = 0xFFFF;
 /// [`PunktfunkPenSample::distance`] sentinel: no hover-distance reading.
 pub const PUNKTFUNK_PEN_DISTANCE_UNKNOWN: u16 = 0xFFFF;
 /// One complete stylus state at one instant ([`punktfunk_connection_send_pen`];
 /// design/pen-tablet-input.md). STATE-FULL, never an edge event: fill every field on every
 /// sample (unknown axes take their `*_UNKNOWN` sentinel) — the host diffs consecutive samples
 /// and synthesizes down/up/button transitions itself, which is what makes a lost datagram
 /// self-heal. `x`/`y` are normalized `0.0..=1.0` in VIDEO-FRAME space (map your letterbox
 /// before filling, exactly like wire touches).
 #[cfg(feature = "quic")]
 #[repr(C)]
 #[derive(Clone, Copy)]
 pub struct PunktfunkPenSample {
    /// Normalized `0.0..=1.0` across the video frame. Must be finite.
    pub x: f32,
    /// Normalized `0.0..=1.0` across the video frame. Must be finite.
    pub y: f32,
    /// Tip force, `0..=65535` full scale (`0` while hovering).
    pub pressure: u16,
    /// Hover distance `0..=65534` (0 = at the hover floor), or `PUNKTFUNK_PEN_DISTANCE_UNKNOWN`.
    pub distance: u16,
    /// Tilt azimuth, degrees `0..=359` clockwise from north, or `PUNKTFUNK_PEN_ANGLE_UNKNOWN`.
    pub azimuth_deg: u16,
    /// Barrel roll (Apple Pencil Pro `rollAngle`), degrees `0..=359`, or
    /// `PUNKTFUNK_PEN_ANGLE_UNKNOWN`.
    pub roll_deg: u16,
    /// µs since the previous sample in the same call (`0` for the first) — the coalesced
    /// capture spacing.
    pub dt_us: u16,
    /// Bitfield of `PUNKTFUNK_PEN_*` state bits. Unknown bits are rejected (`InvalidArg`).
    pub state: u8,
    /// `PUNKTFUNK_PEN_TOOL_PEN` or `PUNKTFUNK_PEN_TOOL_ERASER`.
    pub tool: u8,
    /// Tilt from the surface normal, degrees `0..=90`, or `PUNKTFUNK_PEN_TILT_UNKNOWN`.
    pub tilt_deg: u8,
    /// Set to 0.
    pub _reserved: [u8; 3],
 }
 #[cfg(feature = "quic")]
 impl PunktfunkPenSample {
    /// `None` = invalid field (non-finite coordinate, unknown state bit, unknown tool) —
    /// embedder input is validated strictly, unlike the loss-tolerant wire decode.
    fn to_sample(self) -> Option<crate::quic::PenSample> {
        use crate::quic as q;
        let known = q::PEN_IN_RANGE | q::PEN_TOUCHING | q::PEN_BARREL1 | q::PEN_BARREL2;
        if !self.x.is_finite() || !self.y.is_finite() || self.state & !known != 0 {
            return None;
        }
        let tool = match self.tool {
            PUNKTFUNK_PEN_TOOL_PEN => q::PenTool::Pen,
            PUNKTFUNK_PEN_TOOL_ERASER => q::PenTool::Eraser,
            _ => return None,
        };
        Some(q::PenSample {
            state: self.state,
            tool,
            x: self.x,
            y: self.y,
            pressure: self.pressure,
            distance: self.distance,
            tilt_deg: self.tilt_deg,
            azimuth_deg: self.azimuth_deg,
            roll_deg: self.roll_deg,
            dt_us: self.dt_us,
        })
    }
 }
 /// Read an optional NUL-terminated UTF-8 string parameter; `Err` = invalid pointer/UTF-8.
 #[cfg(feature = "quic")]
 unsafe fn opt_cstr<'a>(p: *const std::os::raw::c_char) -> std::result::Result<Option<&'a str>, ()> {
@@ -985,6 +1076,12 @@ pub const PUNKTFUNK_HOST_CAP_GAMEPAD_STATE: u8 = 0x01;
 /// Host-capability bit in [`punktfunk_connection_host_caps`]: the host supports the shared
 /// clipboard, so a client may offer the toggle. (Mirrors `quic::HOST_CAP_CLIPBOARD`.)
 pub const PUNKTFUNK_HOST_CAP_CLIPBOARD: u8 = 0x02;
 /// Host-capability bit in [`punktfunk_connection_host_caps`]: the host injects full-fidelity
 /// stylus input, so a capable client splits pen contacts out of its touch path and sends them
 /// via [`punktfunk_connection_send_pen`]; without the bit that call returns `Unsupported` and
 /// the client keeps its pen-as-touch fallback. (Mirrors `quic::HOST_CAP_PEN`;
 /// design/pen-tablet-input.md.)
 pub const PUNKTFUNK_HOST_CAP_PEN: u8 = 0x10;
 // Keep the ABI cap bits in lockstep with the wire constants (compile-time guard against drift).
 #[cfg(feature = "quic")]
@@ -998,6 +1095,15 @@ const _: () = {
    assert!(PUNKTFUNK_CODEC_PYROWAVE == crate::quic::CODEC_PYROWAVE);
    assert!(PUNKTFUNK_HOST_CAP_GAMEPAD_STATE == crate::quic::HOST_CAP_GAMEPAD_STATE);
    assert!(PUNKTFUNK_HOST_CAP_CLIPBOARD == crate::quic::HOST_CAP_CLIPBOARD);
    assert!(PUNKTFUNK_HOST_CAP_PEN == crate::quic::HOST_CAP_PEN);
    assert!(PUNKTFUNK_PEN_IN_RANGE == crate::quic::PEN_IN_RANGE);
    assert!(PUNKTFUNK_PEN_TOUCHING == crate::quic::PEN_TOUCHING);
    assert!(PUNKTFUNK_PEN_BARREL1 == crate::quic::PEN_BARREL1);
    assert!(PUNKTFUNK_PEN_BARREL2 == crate::quic::PEN_BARREL2);
    assert!(PUNKTFUNK_PEN_BATCH_MAX as usize == crate::quic::PEN_BATCH_MAX);
    assert!(PUNKTFUNK_PEN_TILT_UNKNOWN == crate::quic::PEN_TILT_UNKNOWN);
    assert!(PUNKTFUNK_PEN_ANGLE_UNKNOWN == crate::quic::PEN_ANGLE_UNKNOWN);
    assert!(PUNKTFUNK_PEN_DISTANCE_UNKNOWN == crate::quic::PEN_DISTANCE_UNKNOWN);
 };
 // Keep the ABI gamepad constants in lockstep with the wire enum (compile-time guard against drift).
@@ -1400,6 +1506,7 @@ pub unsafe extern "C" fn punktfunk_connect_ex7(
        connect_ex_impl(
            host,
            port,
            0, // pre-v11 variant: no client caps
            width,
            height,
            refresh_hz,
@@ -1459,6 +1566,7 @@ pub unsafe extern "C" fn punktfunk_connect_ex8(
        connect_ex_impl(
            host,
            port,
            0, // pre-v11 variant: no client caps
            width,
            height,
            refresh_hz,
@@ -1480,6 +1588,70 @@ pub unsafe extern "C" fn punktfunk_connect_ex8(
    }
 }
 /// Like [`punktfunk_connect_ex8`], plus `client_caps` (ABI v11): a bitfield of
 /// `PUNKTFUNK_CLIENT_CAP_CURSOR` (0x01). Setting the cursor bit asks the host to STOP
 /// compositing the pointer into the video and forward it out-of-band instead — the embedder
 /// MUST then drain [`punktfunk_connection_next_cursor_shape`] /
 /// [`punktfunk_connection_next_cursor_state`] and draw the pointer itself, or the session has
 /// no visible cursor at all. Pass 0 for the composited behavior of every earlier variant.
 ///
 /// # Safety
 /// Same as [`punktfunk_connect_ex8`].
 #[cfg(feature = "quic")]
 #[no_mangle]
 #[allow(clippy::too_many_arguments)]
 pub unsafe extern "C" fn punktfunk_connect_ex9(
    host: *const std::os::raw::c_char,
    port: u16,
    width: u32,
    height: u32,
    refresh_hz: u32,
    compositor: u32,
    gamepad: u32,
    bitrate_kbps: u32,
    video_caps: u8,
    audio_channels: u8,
    video_codecs: u8,
    preferred_codec: u8,
    client_caps: u8,
    launch_id: *const std::os::raw::c_char,
    pin_sha256: *const u8,
    observed_sha256_out: *mut u8,
    client_cert_pem: *const std::os::raw::c_char,
    client_key_pem: *const std::os::raw::c_char,
    timeout_ms: u32,
    status_out: *mut i32,
 ) -> *mut PunktfunkConnection {
    unsafe {
        connect_ex_impl(
            host,
            port,
            client_caps,
            width,
            height,
            refresh_hz,
            compositor,
            gamepad,
            bitrate_kbps,
            video_caps,
            audio_channels,
            video_codecs,
            preferred_codec,
            launch_id,
            pin_sha256,
            observed_sha256_out,
            client_cert_pem,
            client_key_pem,
            timeout_ms,
            status_out,
        )
    }
 }
 /// [`punktfunk_connect_ex9`] `client_caps` bit: render the host cursor locally (the cursor
 /// channel, `design/remote-desktop-sweep.md` M2).
 pub const PUNKTFUNK_CLIENT_CAP_CURSOR: u8 = 0x01;
 /// Shared body of [`punktfunk_connect_ex7`] / [`punktfunk_connect_ex8`]: `status_out`
 /// (nullable) is written on EVERY path — `Ok`, the mapped [`PunktfunkError`],
 /// `InvalidArg` for bad arguments, `Panic` if the connect panicked.
@@ -1488,6 +1660,7 @@ pub unsafe extern "C" fn punktfunk_connect_ex8(
 unsafe fn connect_ex_impl(
    host: *const std::os::raw::c_char,
    port: u16,
    client_caps: u8,
    width: u32,
    height: u32,
    refresh_hz: u32,
@@ -1574,6 +1747,10 @@ unsafe fn connect_ex_impl(
            // themselves (EDR / MediaCodec), so the host's EDID defaults are fine there. An `ex8`
            // variant can carry it if a passthrough embedder ever needs it.
            None,
            // ABI v11 ([`punktfunk_connect_ex9`]): CLIENT_CAP_CURSOR here asks the host to STOP
            // compositing the pointer — only an embedder that renders the cursor planes
            // ([`punktfunk_connection_next_cursor_shape`]/`_state`) may set it. ex7/ex8 pass 0.
            client_caps,
            launch,
            pin,
            identity,
@@ -1593,6 +1770,7 @@ unsafe fn connect_ex_impl(
                    last_audio: std::sync::Mutex::new(None),
                    audio_pcm: std::sync::Mutex::new(AudioPcmState::default()),
                    last_clip: std::sync::Mutex::new(None),
                    last_cursor_shape: std::sync::Mutex::new(None),
                }))
            }
            Err(e) => {
@@ -2242,6 +2420,151 @@ pub unsafe extern "C" fn punktfunk_connection_next_hdr_meta(
    })
 }
 /// One forwarded host-cursor shape (ABI v11, the cursor channel): straight-alpha RGBA8, no
 /// padding, `len == w * h * 4`, hotspot within `w`×`h`. `serial` is the identity
 /// [`PunktfunkCursorState`] refers to — cache the built OS cursor by it.
 #[repr(C)]
 pub struct PunktfunkCursorShape {
    pub serial: u32,
    pub w: u16,
    pub h: u16,
    pub hot_x: u16,
    pub hot_y: u16,
    /// Borrows connection memory until the NEXT cursor-shape call (the audio contract).
    pub rgba: *const u8,
    pub len: usize,
 }
 /// Per-frame host-cursor state (ABI v11): position (the pointer/hotspot point in the host
 /// video's pixel space), visibility, and the host-driven relative-mode hint. `flags` bit 0 =
 /// visible, bit 1 = relative hint (a host app grabbed/hid the pointer — run captured
 /// relative; clear = return to absolute, reappearing at `x`/`y`).
 #[repr(C)]
 pub struct PunktfunkCursorState {
    pub serial: u32,
    pub flags: u8,
    pub x: i32,
    pub y: i32,
 }
 /// Pull the next forwarded cursor SHAPE (sent on pointer-bitmap change over the reliable
 /// control stream; only a session connected with `PUNKTFUNK_CLIENT_CAP_CURSOR` against a
 /// capable host receives any). On `Ok`, `out->rgba` borrows connection memory until the next
 /// cursor-shape call on this handle. Drain from a dedicated thread (one thread per plane).
 ///
 /// # Safety
 /// `c` is a valid connection handle; `out` is writable. At most one thread pulls cursor
 /// shapes; it may run concurrently with every other plane's puller.
 #[cfg(feature = "quic")]
 #[no_mangle]
 pub unsafe extern "C" fn punktfunk_connection_next_cursor_shape(
    c: *mut PunktfunkConnection,
    out: *mut PunktfunkCursorShape,
    timeout_ms: u32,
 ) -> PunktfunkStatus {
    guard(|| {
        let c = match unsafe { c.as_ref() } {
            Some(c) => c,
            None => return PunktfunkStatus::NullPointer,
        };
        if out.is_null() {
            return PunktfunkStatus::NullPointer;
        }
        match c
            .inner
            .next_cursor_shape(std::time::Duration::from_millis(timeout_ms as u64))
        {
            Ok(shape) => {
                let mut slot = c.last_cursor_shape.lock().unwrap();
                *slot = Some(shape);
                let sh = slot.as_ref().unwrap();
                unsafe {
                    *out = PunktfunkCursorShape {
                        serial: sh.serial,
                        w: sh.w,
                        h: sh.h,
                        hot_x: sh.hot_x,
                        hot_y: sh.hot_y,
                        rgba: sh.rgba.as_ptr(),
                        len: sh.rgba.len(),
                    };
                }
                PunktfunkStatus::Ok
            }
            Err(e) => e.status(),
        }
    })
 }
 /// Pull the next cursor STATE (a `0xD0` datagram per host encode tick — latest-wins; drain
 /// the queue and apply only the newest). Same negotiation gate as
 /// [`punktfunk_connection_next_cursor_shape`].
 ///
 /// # Safety
 /// `c` is a valid connection handle; `out` is writable. At most one thread pulls cursor
 /// state; it may run concurrently with every other plane's puller.
 #[cfg(feature = "quic")]
 #[no_mangle]
 pub unsafe extern "C" fn punktfunk_connection_next_cursor_state(
    c: *mut PunktfunkConnection,
    out: *mut PunktfunkCursorState,
    timeout_ms: u32,
 ) -> PunktfunkStatus {
    guard(|| {
        let c = match unsafe { c.as_ref() } {
            Some(c) => c,
            None => return PunktfunkStatus::NullPointer,
        };
        if out.is_null() {
            return PunktfunkStatus::NullPointer;
        }
        match c
            .inner
            .next_cursor_state(std::time::Duration::from_millis(timeout_ms as u64))
        {
            Ok(st) => {
                unsafe {
                    *out = PunktfunkCursorState {
                        serial: st.serial,
                        flags: st.flags,
                        x: st.x,
                        y: st.y,
                    };
                }
                PunktfunkStatus::Ok
            }
            Err(e) => e.status(),
        }
    })
 }
 /// Tell the host who renders the pointer (design/remote-desktop-sweep.md §8 — the mid-stream
 /// mouse-model flip): `client_draws = true` = this client draws it locally (the desktop mouse
 /// model; the host excludes the pointer from the video and forwards shape/state), `false` =
 /// the host composites it into the video (the capture model — full fidelity, the pre-channel
 /// look). Idempotent, latest-wins; harmless against hosts without the cursor cap (an unknown
 /// control message type, ignored). ABI v12.
 ///
 /// # Safety
 /// `c` is a valid connection handle.
 #[cfg(feature = "quic")]
 #[no_mangle]
 pub unsafe extern "C" fn punktfunk_connection_set_cursor_render(
    c: *mut PunktfunkConnection,
    client_draws: bool,
 ) -> PunktfunkStatus {
    guard(|| {
        let c = match unsafe { c.as_ref() } {
            Some(c) => c,
            None => return PunktfunkStatus::NullPointer,
        };
        match c.inner.set_cursor_render(client_draws) {
            Ok(()) => PunktfunkStatus::Ok,
            Err(e) => e.status(),
        }
    })
 }
 /// Pull the next per-AU host timing (0xCF) into `*out`: the host's capture→sent duration for one
 /// access unit, correlated to the AU by `pts_ns` (see [`PunktfunkHostTiming`]).
 /// [`PunktfunkStatus::NoFrame`] on timeout, [`PunktfunkStatus::Closed`] once the session ended.
@@ -2543,6 +2866,51 @@ pub unsafe extern "C" fn punktfunk_connection_send_rich_input2(
    })
 }
 /// Send one stylus sample batch — `count` (`1..=PUNKTFUNK_PEN_BATCH_MAX`) state-full
 /// [`PunktfunkPenSample`]s, oldest first (a capture callback's coalesced samples) — as one
 /// `0xCC/0x05` pen datagram (non-blocking enqueue; design/pen-tablet-input.md). Split longer
 /// runs into consecutive calls. Gate on `punktfunk_connection_host_caps() &
 /// PUNKTFUNK_HOST_CAP_PEN`: toward a host without the bit this returns
 /// [`PunktfunkStatus::Unsupported`] — keep the pen-as-touch fallback there.
 /// [`PunktfunkStatus::InvalidArg`] on a bad count or a bad sample (non-finite coordinate,
 /// unknown state bit / tool).
 ///
 /// # Safety
 /// `c` is a valid connection handle; `samples` is null or points to `count` valid
 /// [`PunktfunkPenSample`]s.
 #[cfg(feature = "quic")]
 #[no_mangle]
 pub unsafe extern "C" fn punktfunk_connection_send_pen(
    c: *mut PunktfunkConnection,
    samples: *const PunktfunkPenSample,
    count: u32,
 ) -> PunktfunkStatus {
    guard(|| {
        let c = match unsafe { c.as_ref() } {
            Some(c) => c,
            None => return PunktfunkStatus::NullPointer,
        };
        if samples.is_null() {
            return PunktfunkStatus::NullPointer;
        }
        if count == 0 || count > PUNKTFUNK_PEN_BATCH_MAX {
            return PunktfunkStatus::InvalidArg;
        }
        let raw = unsafe { std::slice::from_raw_parts(samples, count as usize) };
        let mut batch = [crate::quic::PenSample::default(); crate::quic::PEN_BATCH_MAX];
        for (slot, s) in batch.iter_mut().zip(raw) {
            match s.to_sample() {
                Some(v) => *slot = v,
                None => return PunktfunkStatus::InvalidArg,
            }
        }
        match c.inner.send_pen(&batch[..count as usize]) {
            Ok(()) => PunktfunkStatus::Ok,
            Err(e) => e.status(),
        }
    })
 }
 /// The currently active session mode — the Welcome's, until an accepted
 /// [`punktfunk_connection_request_mode`] switches it. Safe any time after connect.
 ///
@@ -2757,9 +3125,10 @@ fn build_clip_event(
 }
 /// The host capability bitfield the session's `Welcome` carried — a bitfield of
-/// `PUNKTFUNK_HOST_CAP_GAMEPAD_STATE` / `PUNKTFUNK_HOST_CAP_CLIPBOARD`. A client tests
+/// `PUNKTFUNK_HOST_CAP_GAMEPAD_STATE` / `PUNKTFUNK_HOST_CAP_CLIPBOARD` /
-/// `caps & PUNKTFUNK_HOST_CAP_CLIPBOARD` to decide whether to offer the shared-clipboard toggle.
+/// `PUNKTFUNK_HOST_CAP_PEN`. A client tests `caps & PUNKTFUNK_HOST_CAP_CLIPBOARD` to decide
-/// Safe any time after connect.
+/// whether to offer the shared-clipboard toggle, `caps & PUNKTFUNK_HOST_CAP_PEN` before
 /// sending stylus batches. Safe any time after connect.
 ///
 /// # Safety
 /// `c` is a valid connection handle; `caps` is writable (NULL is skipped).
@@ -28,6 +28,10 @@ pub(crate) enum CtrlRequest {
    /// Announce that the local clipboard changed — the lazy format-list offer (bytes cross later on
    /// a fetch stream). Symmetric message; the host may send one too.
    ClipOffer(ClipOffer),
    /// Who renders the pointer (cursor-forward sessions): `true` = client draws locally (the
    /// desktop mouse model — host excludes + forwards), `false` = host composites into the
    /// video (the capture model). Sent on every mouse-model flip; idempotent, latest-wins.
    CursorRender(crate::quic::CursorRenderMode),
 }
 /// What the worker reports to [`NativeClient::connect`] once the handshake lands: the
@@ -20,7 +20,7 @@ use crate::quic::{
    RfiRequest, RichInput,
 };
 use crate::session::Frame;
-use std::sync::atomic::{AtomicBool, AtomicI64, AtomicU32, AtomicU64, Ordering};
+use std::sync::atomic::{AtomicBool, AtomicI64, AtomicU16, AtomicU32, AtomicU64, Ordering};
 use std::sync::mpsc::{Receiver, RecvTimeoutError};
 use std::sync::{Arc, Mutex};
 use std::time::{Duration, Instant};
@@ -42,8 +42,8 @@ pub use self::rumble::{ActuatorQuirks, RumbleCommand};
 use self::control::{CtrlRequest, Negotiated};
 use self::frame_channel::{DecodeLatAcc, FrameChannel, FramePop};
 use self::planes::{
-    RumbleUpdate, AUDIO_QUEUE, CLIP_EVENT_QUEUE, HDR_META_QUEUE, HIDOUT_QUEUE, HOST_TIMING_QUEUE,
+    RumbleUpdate, AUDIO_QUEUE, CLIP_EVENT_QUEUE, CURSOR_SHAPE_QUEUE, CURSOR_STATE_QUEUE,
-    RUMBLE_QUEUE,
+    HDR_META_QUEUE, HIDOUT_QUEUE, HOST_TIMING_QUEUE, RUMBLE_QUEUE,
 };
 use self::probe::ProbeState;
 use self::pump::run_pump;
@@ -93,13 +93,22 @@ pub struct NativeClient {
    /// Inbound per-AU host capture→send timings — 0xCF datagrams (the client always advertises
    /// [`quic::VIDEO_CAP_HOST_TIMING`]; an older host simply never sends any).
    host_timing: Mutex<Receiver<crate::quic::HostTiming>>,
    /// Inbound cursor shapes (control-stream [`crate::quic::CursorShape`]) — only a session
    /// that advertised [`quic::CLIENT_CAP_CURSOR`] against a [`quic::HOST_CAP_CURSOR`] host
    /// ever receives any.
    cursor_shape: Mutex<Receiver<crate::quic::CursorShape>>,
    /// Inbound per-frame cursor state — `0xD0` datagrams (same negotiation gate as shapes).
    cursor_state: Mutex<Receiver<crate::quic::CursorState>>,
    input_tx: tokio::sync::mpsc::UnboundedSender<InputEvent>,
    /// Outbound mic frames `(seq, pts_ns, opus)` → encoded as 0xCB datagrams by the worker.
    /// Bounded ([`MIC_QUEUE`]): a wedged worker drops fresh frames (logged) instead of queueing
    /// audio-latency (and memory) without limit — mic is best-effort end to end.
    mic_tx: tokio::sync::mpsc::Sender<(u32, u64, Vec<u8>)>,
-    /// Outbound rich input (DualSense touchpad / motion) → 0xCC datagrams by the worker.
+    /// Outbound 0xCC rich-input plane, PRE-ENCODED datagrams: [`RichInput`] touchpad/motion
-    rich_input_tx: tokio::sync::mpsc::UnboundedSender<RichInput>,
+    /// (encoded in [`NativeClient::send_rich_input`]) and stylus [`crate::quic::PenBatch`]es
    /// (encoded in [`NativeClient::send_pen`]) share the channel — the worker's task just
    /// forwards bytes, so a new 0xCC kind never touches the pump.
    rich_input_tx: tokio::sync::mpsc::UnboundedSender<Vec<u8>>,
    /// Outbound control-stream requests (mode switch, speed test) → the worker's control task.
    /// Bounded ([`CTRL_QUEUE`]) — the requests are sparse; a full queue means the control task
    /// is wedged/dead, and callers treat it like a closed session.
@@ -115,6 +124,9 @@ pub struct NativeClient {
    /// Monotonic id for outbound fetches ([`NativeClient::clip_fetch`]); stays below
    /// [`crate::clipboard::INBOUND_REQ_FLAG`] so it never collides with an inbound serve `req_id`.
    next_xfer_id: AtomicU32,
    /// Wrapping per-connection [`crate::quic::PenBatch::seq`] counter, stamped by
    /// [`NativeClient::send_pen`] (the host's reorder gate compares it).
    pen_seq: AtomicU16,
    /// The host capability bitfield ([`crate::quic::Welcome::host_caps`]) — see
    /// [`NativeClient::host_caps`].
    pub host_caps: u8,
@@ -316,6 +328,12 @@ impl NativeClient {
        // display's EDID so host apps tone-map to the client's real panel; `None` = unknown/SDR
        // (the host keeps its built-in EDID defaults). See [`crate::quic::Hello::display_hdr`].
        display_hdr: Option<HdrMeta>,
        // Non-video client capabilities ([`crate::quic::Hello::client_caps`]) — set
        // [`crate::quic::CLIENT_CAP_CURSOR`] ONLY if this embedder actually renders the host
        // cursor locally (shape + state planes): the host stops compositing the pointer into
        // the video for a session that advertises it, so a non-rendering embedder that sets it
        // streams with NO visible cursor at all. `0` = today's composited behavior.
        client_caps: u8,
        launch: Option<String>,
        pin: Option<[u8; 32]>,
        identity: Option<(String, String)>,
@@ -332,11 +350,15 @@ impl NativeClient {
            std::sync::mpsc::sync_channel::<crate::quic::HostTiming>(HOST_TIMING_QUEUE);
        let (input_tx, input_rx) = tokio::sync::mpsc::unbounded_channel::<InputEvent>();
        let (mic_tx, mic_rx) = tokio::sync::mpsc::channel::<(u32, u64, Vec<u8>)>(MIC_QUEUE);
-        let (rich_input_tx, rich_input_rx) = tokio::sync::mpsc::unbounded_channel::<RichInput>();
+        let (rich_input_tx, rich_input_rx) = tokio::sync::mpsc::unbounded_channel::<Vec<u8>>();
        let (ctrl_tx, ctrl_rx) = tokio::sync::mpsc::channel::<CtrlRequest>(CTRL_QUEUE);
        let (clip_event_tx, clip_event_rx) =
            std::sync::mpsc::sync_channel::<ClipEventCore>(CLIP_EVENT_QUEUE);
        let (clip_cmd_tx, clip_cmd_rx) = tokio::sync::mpsc::unbounded_channel::<ClipCommand>();
        let (cursor_shape_tx, cursor_shape_rx) =
            std::sync::mpsc::sync_channel::<crate::quic::CursorShape>(CURSOR_SHAPE_QUEUE);
        let (cursor_state_tx, cursor_state_rx) =
            std::sync::mpsc::sync_channel::<crate::quic::CursorState>(CURSOR_STATE_QUEUE);
        let (ready_tx, ready_rx) = std::sync::mpsc::channel::<Result<Negotiated>>();
        let shutdown = Arc::new(AtomicBool::new(false));
        let quit = Arc::new(AtomicBool::new(false));
@@ -390,6 +412,7 @@ impl NativeClient {
                    video_codecs,
                    preferred_codec,
                    display_hdr,
                    client_caps,
                    launch,
                    pin,
                    identity,
@@ -401,6 +424,8 @@ impl NativeClient {
                    hidout_tx,
                    hdr_meta_tx,
                    host_timing_tx,
                    cursor_shape_tx,
                    cursor_state_tx,
                    input_rx,
                    mic_rx,
                    rich_input_rx,
@@ -445,6 +470,8 @@ impl NativeClient {
            hidout: Mutex::new(hidout_rx),
            hdr_meta: Mutex::new(hdr_meta_rx),
            host_timing: Mutex::new(host_timing_rx),
            cursor_shape: Mutex::new(cursor_shape_rx),
            cursor_state: Mutex::new(cursor_state_rx),
            input_tx,
            mic_tx,
            rich_input_tx,
@@ -452,6 +479,7 @@ impl NativeClient {
            clip: Mutex::new(clip_event_rx),
            clip_cmd_tx,
            next_xfer_id: AtomicU32::new(1),
            pen_seq: AtomicU16::new(0),
            host_caps: negotiated.host_caps,
            probe,
            shutdown,
@@ -552,6 +580,20 @@ impl NativeClient {
            .map_err(|_| PunktfunkError::Closed)
    }
    /// Tell the host who renders the pointer (cursor-forward sessions —
    /// design/remote-desktop-sweep.md §8): `true` = this client draws it locally (the desktop
    /// mouse model; the host excludes the pointer from the video and forwards shape/state),
    /// `false` = the host composites it into the video (the capture model — full fidelity,
    /// the pre-channel behavior). Call on every mouse-model flip; idempotent, latest-wins,
    /// no-op on hosts without [`HOST_CAP_CURSOR`](crate::quic::HOST_CAP_CURSOR).
    pub fn set_cursor_render(&self, client_draws: bool) -> Result<()> {
        self.ctrl_tx
            .try_send(CtrlRequest::CursorRender(crate::quic::CursorRenderMode {
                client_draws,
            }))
            .map_err(|_| PunktfunkError::Closed)
    }
    /// Ask the host's encoder to emit a fresh IDR keyframe now (client recovery on a stalled
    /// decode). Non-blocking, fire-and-forget — the recovered keyframe is the only ack. The
    /// caller should throttle (the decode stays wedged across several frames until the IDR
@@ -892,6 +934,32 @@ impl NativeClient {
        }
    }
    /// Pull the next host cursor shape (design/remote-desktop-sweep.md M2): RGBA bitmap +
    /// hotspot, sent on pointer-bitmap change over the reliable control stream. The embedder
    /// caches by `serial` and builds an OS cursor from it; [`NativeClient::next_cursor_state`]
    /// references shapes by serial. Only a session that advertised
    /// [`crate::quic::CLIENT_CAP_CURSOR`] against a capable host receives any. Same
    /// timeout/closed semantics as [`NativeClient::next_hidout`].
    pub fn next_cursor_shape(&self, timeout: Duration) -> Result<crate::quic::CursorShape> {
        match self.cursor_shape.lock().unwrap().recv_timeout(timeout) {
            Ok(s) => Ok(s),
            Err(RecvTimeoutError::Timeout) => Err(PunktfunkError::NoFrame),
            Err(RecvTimeoutError::Disconnected) => Err(PunktfunkError::Closed),
        }
    }
    /// Pull the next per-frame cursor state (`0xD0`): position, visibility and the M3
    /// relative-mode hint, referencing a shape by serial. Latest-wins — an embedder should
    /// drain the queue and apply only the newest. Same negotiation gate and timeout/closed
    /// semantics as [`NativeClient::next_cursor_shape`].
    pub fn next_cursor_state(&self, timeout: Duration) -> Result<crate::quic::CursorState> {
        match self.cursor_state.lock().unwrap().recv_timeout(timeout) {
            Ok(s) => Ok(s),
            Err(RecvTimeoutError::Timeout) => Err(PunktfunkError::NoFrame),
            Err(RecvTimeoutError::Disconnected) => Err(PunktfunkError::Closed),
        }
    }
    /// Pull the next per-AU host timing (0xCF): the host's capture→sent duration for one access
    /// unit, correlated to the AU by `pts_ns`. Feeds the unified stats HUD's `host` / `network`
    /// split (`network = (received + clock_offset − pts) − host_us`); a stats consumer should
@@ -1012,7 +1080,39 @@ impl NativeClient {
    /// loss like every datagram. No-op unless the host runs the DualSense gamepad backend.
    pub fn send_rich_input(&self, rich: RichInput) -> Result<()> {
        self.rich_input_tx
-            .send(rich)
+            .send(rich.encode())
            .map_err(|_| PunktfunkError::Closed)
    }
    /// Queue one stylus sample batch for delivery as a `0xCC/0x05` pen datagram
    /// (design/pen-tablet-input.md). `samples` are state-full and oldest-first (a capture
    /// callback's coalesced samples), at most [`crate::quic::PEN_BATCH_MAX`] per call — split
    /// longer runs into consecutive calls so the stamped wrapping `seq` keeps them ordered.
    /// Best-effort like every datagram: a lost batch self-heals on the next one (the samples
    /// carry full state, the host diffs — see [`crate::quic::PenTracker`]).
    ///
    /// **Heartbeat contract**: while the pen is in range or touching, repeat the last sample
    /// at least every ~100 ms even when nothing changed (capture APIs are silent for a
    /// stationary pen) — the host force-releases the stroke after
    /// [`crate::quic::PEN_TOUCH_TIMEOUT_MS`] of silence as its dead-client failsafe.
    ///
    /// Requires the host to have advertised [`crate::quic::HOST_CAP_PEN`]; toward an older
    /// host this returns `Unsupported` (embedders keep their pen-as-touch fallback instead of
    /// spraying 240 Hz datagrams the host drops unread).
    pub fn send_pen(&self, samples: &[crate::quic::PenSample]) -> Result<()> {
        if self.host_caps & crate::quic::HOST_CAP_PEN == 0 {
            return Err(PunktfunkError::Unsupported(
                "host did not advertise HOST_CAP_PEN",
            ));
        }
        if samples.is_empty() || samples.len() > crate::quic::PEN_BATCH_MAX {
            return Err(PunktfunkError::InvalidArg(
                "pen batch must hold 1..=PEN_BATCH_MAX samples",
            ));
        }
        let seq = self.pen_seq.fetch_add(1, Ordering::Relaxed);
        self.rich_input_tx
            .send(crate::quic::PenBatch::new(seq, samples).encode())
            .map_err(|_| PunktfunkError::Closed)
    }
@@ -86,10 +86,22 @@ impl NativeClient {
                    // A typed application close from the host (pairing not armed / armed for a
                    // different device / rate-limited / version mismatch) beats the generic
                    // transport error the aborted exchange produced — it is the actual answer.
-                    Err(e) => Err(match reject_from_close(&conn) {
+                    // Same close-vs-stream-error race as the connect handshake: give the
                    // host's CONNECTION_CLOSE a short grace to be processed before deciding
                    // the error was plain transport trouble.
                    Err(e) => {
                        if conn.close_reason().is_none() {
                            let _ = tokio::time::timeout(
                                std::time::Duration::from_millis(300),
                                conn.closed(),
                            )
                            .await;
                        }
                        Err(match reject_from_close(&conn) {
                            Some(r) => PunktfunkError::Rejected(r),
                            None => e,
-                    }),
+                        })
                    }
                    ok => ok,
                };
                // Always tell the host we're done so it never blocks at its read — code 0 on
@@ -35,6 +35,16 @@ pub(crate) const HOST_TIMING_QUEUE: usize = 512;
 /// a dropped fetch-request makes the serving stream time out and reset cleanly.
 pub(crate) const CLIP_EVENT_QUEUE: usize = 32;
 /// Cursor-shape plane depth (control-stream [`crate::quic::CursorShape`], one per pointer-bitmap
 /// change — human-paced). Overflow drops the newest (try_send); the next shape change or a
 /// serial mismatch against `0xD0` state heals it visually within a shape-change period.
 pub(crate) const CURSOR_SHAPE_QUEUE: usize = 8;
 /// Cursor-state plane depth (`0xD0`, one datagram per captured frame). Latest-wins state — the
 /// embedder drains per present; a tiny ring only bridges scheduling jitter. Overflow drops the
 /// newest (try_send), healed by the very next frame's datagram.
 pub(crate) const CURSOR_STATE_QUEUE: usize = 8;
 /// One Opus packet from the host's audio datagram stream (48 kHz stereo, 5 ms frames).
 #[derive(Clone, Debug)]
 pub struct AudioPacket {
@@ -51,6 +51,8 @@ pub(super) async fn run_pump(args: WorkerArgs) {
        hidout_tx,
        hdr_meta_tx,
        host_timing_tx,
        cursor_shape_tx,
        cursor_state_tx,
        input_rx,
        mut mic_rx,
        mut rich_input_rx,
@@ -97,11 +99,12 @@ pub(super) async fn run_pump(args: WorkerArgs) {
        }
    });
-    // Rich-input task: embedder DualSense touchpad / motion → 0xCC uplink datagrams.
+    // Rich-input task: pre-encoded 0xCC uplink datagrams (DualSense touchpad / motion, pen
    // batches — encoded at the NativeClient surface so new plane kinds never touch the pump).
    let rich_conn = conn.clone();
    tokio::spawn(async move {
-        while let Some(rich) = rich_input_rx.recv().await {
+        while let Some(d) = rich_input_rx.recv().await {
-            let _ = rich_conn.send_datagram(rich.encode().into());
+            let _ = rich_conn.send_datagram(d.into());
        }
    });
@@ -123,6 +126,7 @@ pub(super) async fn run_pump(args: WorkerArgs) {
            clock_offset: clock_offset.clone(),
            clock_gen: clock_gen.clone(),
            clip_event_tx: clip_event_tx.clone(),
            cursor_shape_tx,
        }
        .run(),
    );
@@ -136,6 +140,7 @@ pub(super) async fn run_pump(args: WorkerArgs) {
        hidout_tx,
        hdr_meta_tx,
        host_timing_tx,
        cursor_state_tx,
    ));
    // Clipboard task: the fetch-stream accept loop (host pulls what we offered) + outbound fetches
@@ -21,6 +21,9 @@ pub(super) struct ControlTask {
    /// Clipboard metadata events (ClipState/ClipOffer) feed the same event plane the
    /// clipboard task uses for fetch data.
    pub(super) clip_event_tx: std::sync::mpsc::SyncSender<ClipEventCore>,
    /// Host cursor shapes ([`CursorShape`], sent on pointer-bitmap change) → the embedder's
    /// shape plane ([`NativeClient::next_cursor_shape`]).
    pub(super) cursor_shape_tx: std::sync::mpsc::SyncSender<crate::quic::CursorShape>,
 }
 impl ControlTask {
@@ -36,6 +39,7 @@ impl ControlTask {
            clock_offset,
            clock_gen,
            clip_event_tx,
            cursor_shape_tx,
        } = self;
        // Mid-stream clock re-sync (see [`ClockResync`]): a batch runs every
        // CLOCK_RESYNC_INTERVAL and whenever the pump asks (CtrlRequest::ClockResync after
@@ -67,6 +71,7 @@ impl ControlTask {
                        }
                        CtrlRequest::ClipControl(c) => c.encode(),
                        CtrlRequest::ClipOffer(o) => o.encode(),
                        CtrlRequest::CursorRender(m) => m.encode(),
                    };
                    if io::write_msg(&mut ctrl_send, &bytes).await.is_err() {
                        break;
@@ -167,6 +172,10 @@ impl ControlTask {
                            seq: offer.seq,
                            kinds: offer.kinds,
                        });
                    } else if let Ok(shape) = crate::quic::CursorShape::decode(&msg) {
                        // Pointer bitmap changed (cursor channel, only when negotiated). try_send:
                        // an overflowing ring drops the newest shape — the next change resends.
                        let _ = cursor_shape_tx.try_send(shape);
                    } else {
                        tracing::warn!(
                            tag = ?msg.first(),
@@ -3,6 +3,9 @@
 use super::*;
 // One parameter per demuxed plane — grouping them into a struct would just move the field
 // list one hop away from the single call site.
 #[allow(clippy::too_many_arguments)]
 pub(super) async fn run(
    conn: quinn::Connection,
    audio_tx: std::sync::mpsc::SyncSender<AudioPacket>,
@@ -11,6 +14,7 @@ pub(super) async fn run(
    hidout_tx: std::sync::mpsc::SyncSender<crate::quic::HidOutput>,
    hdr_meta_tx: std::sync::mpsc::SyncSender<crate::quic::HdrMeta>,
    host_timing_tx: std::sync::mpsc::SyncSender<crate::quic::HostTiming>,
    cursor_state_tx: std::sync::mpsc::SyncSender<crate::quic::CursorState>,
 ) {
    // Per-pad reorder gate for v2 rumble envelopes (the seq analog of the host's gamepad-state
    // gate): a datagram the network reordered must not roll a stopped motor back on. Legacy v1
@@ -73,6 +77,11 @@ pub(super) async fn run(
                    let _ = host_timing_tx.try_send(t);
                }
            }
            Some(&crate::quic::CURSOR_STATE_MAGIC) => {
                if let Some(s) = crate::quic::decode_cursor_state_datagram(&d) {
                    let _ = cursor_state_tx.try_send(s);
                }
            }
            _ => {} // unknown tag — a newer host; ignore
        }
    }
@@ -143,6 +143,10 @@ pub(super) async fn connect_and_handshake(args: &WorkerArgs) -> Result<Handshake
                // The client display's HDR volume → the host's virtual-display EDID (host apps
                // tone-map to the client's real panel). `None` = unknown/SDR.
                display_hdr,
                // NOT unconditional like HOST_TIMING above: CLIENT_CAP_CURSOR makes the host
                // stop compositing the pointer, so only an embedder that actually renders the
                // cursor locally may set it (the embedder decides, we pass through).
                client_caps: args.client_caps,
            }
            .encode(),
        )
@@ -240,9 +244,22 @@ pub(super) async fn connect_and_handshake(args: &WorkerArgs) -> Result<Handshake
            negotiated,
            host_caps,
        }),
-        Err(e) => Err(match reject_from_close(&conn) {
+        Err(e) => {
            // The host's typed close can land a beat AFTER the stream error it caused: the
            // stream reset/FIN and the CONNECTION_CLOSE are in flight together, and quinn can
            // hand the reader its mid-frame EOF before it processes the close. Give the close a
            // short grace to arrive so a host-side setup failure renders as its real reason
            // ("the host could not start the stream session") instead of "control stream
            // finished mid-frame". No-op when the connection already closed (or never will —
            // bounded by the timeout).
            if conn.close_reason().is_none() {
                let _ = tokio::time::timeout(std::time::Duration::from_millis(300), conn.closed())
                    .await;
            }
            Err(match reject_from_close(&conn) {
                Some(r) => PunktfunkError::Rejected(r),
                None => e,
-        }),
+            })
        }
    }
 }
@@ -5,7 +5,7 @@ use crate::clipboard::{ClipCommand, ClipEventCore};
 use crate::config::{CompositorPref, GamepadPref, Mode};
 use crate::error::Result;
 use crate::input::InputEvent;
-use crate::quic::{HdrMeta, HidOutput, RichInput};
+use crate::quic::{HdrMeta, HidOutput};
 use std::sync::atomic::{AtomicBool, AtomicI64, AtomicU64};
 use std::sync::mpsc::SyncSender;
 use std::sync::{Arc, Mutex};
@@ -22,6 +22,7 @@ pub(crate) struct WorkerArgs {
    pub(crate) video_codecs: u8,
    pub(crate) preferred_codec: u8,
    pub(crate) display_hdr: Option<HdrMeta>,
    pub(crate) client_caps: u8,
    pub(crate) launch: Option<String>,
    pub(crate) pin: Option<[u8; 32]>,
    pub(crate) identity: Option<(String, String)>,
@@ -38,9 +39,12 @@ pub(crate) struct WorkerArgs {
    pub(crate) hidout_tx: SyncSender<HidOutput>,
    pub(crate) hdr_meta_tx: SyncSender<HdrMeta>,
    pub(crate) host_timing_tx: SyncSender<crate::quic::HostTiming>,
    pub(crate) cursor_shape_tx: SyncSender<crate::quic::CursorShape>,
    pub(crate) cursor_state_tx: SyncSender<crate::quic::CursorState>,
    pub(crate) input_rx: tokio::sync::mpsc::UnboundedReceiver<InputEvent>,
    pub(crate) mic_rx: tokio::sync::mpsc::Receiver<(u32, u64, Vec<u8>)>,
-    pub(crate) rich_input_rx: tokio::sync::mpsc::UnboundedReceiver<RichInput>,
+    /// Pre-encoded 0xCC datagrams (rich input AND pen batches — see `NativeClient.rich_input_tx`).
    pub(crate) rich_input_rx: tokio::sync::mpsc::UnboundedReceiver<Vec<u8>>,
    pub(crate) ctrl_rx: tokio::sync::mpsc::Receiver<CtrlRequest>,
    pub(crate) ctrl_tx: tokio::sync::mpsc::Sender<CtrlRequest>,
    /// Inbound clipboard event plane feed — the control task pushes ClipState/ClipOffer, the
@@ -61,6 +61,7 @@ pub enum PunktfunkStatus {
    RejectedSuperseded = -26,
    RejectedWireVersion = -27,
    RejectedBusy = -28,
    RejectedSetupFailed = -29,
    Panic = -99,
 }
@@ -89,6 +90,7 @@ impl PunktfunkError {
                    R::Superseded => PunktfunkStatus::RejectedSuperseded,
                    R::WireVersionMismatch => PunktfunkStatus::RejectedWireVersion,
                    R::Busy => PunktfunkStatus::RejectedBusy,
                    R::SetupFailed => PunktfunkStatus::RejectedSetupFailed,
                }
            }
        }
@@ -73,6 +73,17 @@ pub enum InputKind {
    /// [`HOST_CAP_GAMEPAD_STATE`](crate::quic::HOST_CAP_GAMEPAD_STATE); an older host ignores the
    /// unknown tag (every pad then uses the session-default kind — the pre-existing behaviour).
    GamepadArrival = 14,
    /// One Unicode scalar of **committed text** — `code` = the scalar value, everything else 0.
    ///
    /// The IME path: the layout-independent VK key events cannot express text an input method
    /// *commits* (autocorrect, gesture typing, non-Latin scripts, emoji), so a capable client
    /// sends the committed characters verbatim and the host injects them directly (Windows
    /// `KEYEVENTF_UNICODE`; Linux wlroots via a dynamically-grown Unicode keymap on a dedicated
    /// virtual keyboard). A multi-character commit is consecutive events in order. Sent only when
    /// the host advertised [`HOST_CAP_TEXT_INPUT`](crate::quic::HOST_CAP_TEXT_INPUT) — toward an
    /// older host (or one whose inject backend can't type text) clients keep the best-effort VK
    /// synthesis, and an older host ignores the unknown tag entirely.
    TextInput = 15,
 }
 /// Pack a [`InputKind::GamepadRemove`] `flags` word (`seq << 24 | pad`) — the same low-byte-pad /
@@ -158,6 +169,7 @@ impl InputKind {
            12 => GamepadState,
            13 => GamepadRemove,
            14 => GamepadArrival,
            15 => TextInput,
            _ => return None,
        })
    }
@@ -392,10 +404,27 @@ mod tests {
            };
            assert_eq!(InputEvent::decode(&e.encode()), Some(e));
        }
-        // GamepadRemove + GamepadArrival are valid kinds; 15 (one past them) is not.
+        // GamepadRemove/GamepadArrival/TextInput are valid kinds; 16 (one past them) is not.
        assert_eq!(InputKind::from_u8(13), Some(InputKind::GamepadRemove));
        assert_eq!(InputKind::from_u8(14), Some(InputKind::GamepadArrival));
-        assert_eq!(InputKind::from_u8(15), None);
+        assert_eq!(InputKind::from_u8(15), Some(InputKind::TextInput));
        assert_eq!(InputKind::from_u8(16), None);
    }
    #[test]
    fn text_input_roundtrip() {
        // One Unicode scalar per event — BMP and astral (emoji) alike.
        for cp in ['a' as u32, 'ß' as u32, '語' as u32, 0x1F600 /* 😀 */] {
            let e = InputEvent {
                kind: InputKind::TextInput,
                _pad: [0; 3],
                code: cp,
                x: 0,
                y: 0,
                flags: 0,
            };
            assert_eq!(InputEvent::decode(&e.encode()), Some(e));
        }
    }
    #[test]
@@ -102,7 +102,14 @@ pub use stats::Stats;
 /// v10: added `punktfunk_connection_clock_offset_now_ns` — the LIVE (mid-stream re-synced)
 /// clock offset ongoing latency math must use; the connect-time getter stays frozen by
 /// contract. Additive, client-local — no wire change, so [`WIRE_VERSION`] is unchanged.
-pub const ABI_VERSION: u32 = 10;
+/// v12: added `punktfunk_connection_set_cursor_render` — the mid-stream cursor-render flip
 /// (design/remote-desktop-sweep.md §8): the client's mouse-model chord tells the host who
 /// renders the pointer. Additive; rides the existing control stream (a new message TYPE, which
 /// pre-§8 hosts ignore), so [`WIRE_VERSION`] is unchanged.
 /// v13: added `punktfunk_connection_send_pen` — the stylus wire plane
 /// (design/pen-tablet-input.md): a client sends `RICH_PEN` sample batches once the host
 /// advertises `HOST_CAP_PEN`. Additive and capability-gated, so [`WIRE_VERSION`] is unchanged.
 pub const ABI_VERSION: u32 = 13;
 /// The punktfunk/1 **wire** version — what `Hello`/`Welcome` carry and hosts equality-check.
 /// Deliberately its own constant: [`ABI_VERSION`] tracks the embeddable **C surface**
@@ -71,6 +71,47 @@ pub const HOST_CAP_GAMEPAD_STATE: u8 = 0x01;
 /// trailing `host_caps` byte — no wire-layout change.
 pub const HOST_CAP_CLIPBOARD: u8 = 0x02;
 /// [`Welcome::host_caps`] bit: the host's active inject backend can type **committed text**
 /// ([`InputKind::TextInput`](crate::input::InputKind::TextInput) — one Unicode scalar per event):
 /// Windows (`KEYEVENTF_UNICODE`) and Linux wlroots (dynamic Unicode keymap on a dedicated virtual
 /// keyboard); the KWin/libei/gamescope backends can only press layout keycodes, so those sessions
 /// don't set it. A capable client routes its IME's committed text (autocorrect, gesture typing,
 /// non-Latin scripts, emoji) through `TextInput` instead of lossy VK synthesis; absent the bit it
 /// keeps the VK fallback. Packs into the existing trailing `host_caps` byte — no wire-layout
 /// change; an older host ignores the unknown input tag anyway (input is lossy by design).
 pub const HOST_CAP_TEXT_INPUT: u8 = 0x04;
 /// [`Hello::client_caps`] bit: the client renders the host cursor LOCALLY
 /// (design/remote-desktop-sweep.md M2). It consumes [`CursorShape`](super::control::CursorShape)
 /// control messages (RGBA bitmap + hotspot, cached by serial) and per-frame
 /// [`CursorState`](super::datagram::CursorState) `0xD0` datagrams (position/visibility), and
 /// draws the pointer itself — so the host must STOP compositing the cursor into the video
 /// (`SessionPlan.cursor_blend = false`) or the user sees it twice. Active only when the host
 /// answers with [`HOST_CAP_CURSOR`] (capable-and-agreed, the 444/clipboard precedent); toward
 /// an older or incapable host nothing changes.
 pub const CLIENT_CAP_CURSOR: u8 = 0x01;
 /// [`Welcome::host_caps`] bit: the host CAN forward the cursor out-of-band (it captures cursor
 /// metadata separately from the frame — the Linux portal `SPA_META_Cursor` path; NOT gamescope,
 /// whose capture carries no cursor, and NOT Windows yet, where DWM composites into the IDD
 /// frame). Set only when the client asked via [`CLIENT_CAP_CURSOR`]; when both bits agree the
 /// host stops blending and ships [`CursorShape`](super::control::CursorShape) +
 /// [`CursorState`](super::datagram::CursorState) instead. `0x08` — `0x04` is
 /// [`HOST_CAP_TEXT_INPUT`], `0x01`/`0x02` are gamepad-state / clipboard.
 pub const HOST_CAP_CURSOR: u8 = 0x08;
 /// [`Welcome::host_caps`] bit: the host injects full-fidelity stylus input — it routes
 /// [`PenBatch`](super::pen::PenBatch) `0xCC/0x05` datagrams (pressure, tilt, azimuth, barrel
 /// roll, hover, eraser, barrel buttons) through the [`PenTracker`](super::pen::PenTracker)
 /// into a virtual tablet device (design/pen-tablet-input.md). A capable client (Apple Pencil,
 /// Android stylus) then splits pen contacts out of its finger/touch path and sends pen
 /// batches; absent the bit it keeps folding the pen into touch/pointer like today, and
 /// [`NativeClient::send_pen`](crate::client::NativeClient::send_pen) refuses to send. The
 /// wire ships ahead of the backend (P0): no host sets this bit until the P1 injector lands —
 /// which is exactly why the gate exists. `0x10` — `0x08` is [`HOST_CAP_CURSOR`], `0x04` is
 /// [`HOST_CAP_TEXT_INPUT`], `0x01`/`0x02` are gamepad-state / clipboard.
 pub const HOST_CAP_PEN: u8 = 0x10;
 /// [`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.
@@ -783,11 +783,146 @@ impl ClipFetchHdr {
    }
 }
 // --- Cursor channel (design/remote-desktop-sweep.md M2) --------------------------------------
 // The host cursor, forwarded out-of-band so the CLIENT draws it as a real OS cursor (the
 // Parsec/RDP model) instead of paying the video round-trip. Shape (rare, needs reliability)
 // rides here on the control stream; per-frame position/visibility rides the lossy `0xD0`
 // datagram plane ([`super::datagram::CursorState`]). Active only when the client's
 // [`CLIENT_CAP_CURSOR`](super::caps::CLIENT_CAP_CURSOR) met the host's
 // [`HOST_CAP_CURSOR`](super::caps::HOST_CAP_CURSOR) — the host stops compositing then.
 // ---------------------------------------------------------------------------------------------
 /// Type byte of [`CursorShape`] (host → client): the pointer's bitmap + hotspot changed.
 pub const MSG_CURSOR_SHAPE: u8 = 0x50;
 /// Type byte of [`CursorRenderMode`] (client → host): who renders the pointer right now.
 pub const MSG_CURSOR_RENDER: u8 = 0x51;
 /// Per-side pixel cap for a forwarded cursor bitmap. The control-stream frame is length-prefixed
 /// with a `u16`, so a whole message must fit 65535 bytes — 128×128 RGBA (65536 B) already
 /// overshoots before the 17-byte header. 120² (57.6 KiB + header) fits with headroom and covers
 /// real cursors (typically ≤ 64 px, ≤ 96 px at HiDPI scale); the HOST downscales anything
 /// larger before forwarding, so the cap is invisible to clients.
 pub const CURSOR_SHAPE_MAX_SIDE: u16 = 120;
 /// `host → client` ([`MSG_CURSOR_SHAPE`]): one cursor shape, sent when the pointer's bitmap
 /// changes (never per-frame — [`super::datagram::CursorState`] carries the motion). The client
 /// caches shapes by `serial` and re-installs a cached one without any bitmap crossing again
 /// (the RDP pointer-cache idea for free: re-showing a known serial is a 14-byte
 /// [`super::datagram::CursorState`], not a resend).
 #[derive(Clone, Debug, PartialEq, Eq)]
 pub struct CursorShape {
    /// Bitmap identity — bumped by the host's capture layer only on shape change; position
    /// moves keep the serial stable. [`super::datagram::CursorState::serial`] references it.
    pub serial: u32,
    /// Bitmap dimensions in pixels, `1..=`[`CURSOR_SHAPE_MAX_SIDE`] each.
    pub w: u16,
    pub h: u16,
    /// Hotspot (the pixel that IS the pointer position), within `w`×`h`.
    pub hot_x: u16,
    pub hot_y: u16,
    /// Straight-alpha RGBA8, exactly `w * h * 4` bytes, no padding.
    pub rgba: Vec<u8>,
 }
 impl CursorShape {
    pub fn encode(&self) -> Vec<u8> {
        // magic[0..4] type[4] serial[5..9] w[9..11] h[11..13] hot_x[13..15] hot_y[15..17] rgba…
        let mut b = Vec::with_capacity(17 + self.rgba.len());
        b.extend_from_slice(CTL_MAGIC);
        b.push(MSG_CURSOR_SHAPE);
        b.extend_from_slice(&self.serial.to_le_bytes());
        b.extend_from_slice(&self.w.to_le_bytes());
        b.extend_from_slice(&self.h.to_le_bytes());
        b.extend_from_slice(&self.hot_x.to_le_bytes());
        b.extend_from_slice(&self.hot_y.to_le_bytes());
        b.extend_from_slice(&self.rgba);
        b
    }
    pub fn decode(b: &[u8]) -> Result<CursorShape> {
        if b.len() < 17 || &b[0..4] != CTL_MAGIC || b[4] != MSG_CURSOR_SHAPE {
            return Err(PunktfunkError::InvalidArg("bad CursorShape"));
        }
        let u16at = |o: usize| u16::from_le_bytes([b[o], b[o + 1]]);
        let (w, h) = (u16at(9), u16at(11));
        if w == 0 || h == 0 || w > CURSOR_SHAPE_MAX_SIDE || h > CURSOR_SHAPE_MAX_SIDE {
            return Err(PunktfunkError::InvalidArg("bad CursorShape dims"));
        }
        if b.len() != 17 + (w as usize) * (h as usize) * 4 {
            return Err(PunktfunkError::InvalidArg("bad CursorShape len"));
        }
        Ok(CursorShape {
            serial: u32::from_le_bytes(b[5..9].try_into().unwrap()),
            w,
            h,
            hot_x: u16at(13),
            hot_y: u16at(15),
            rgba: b[17..].to_vec(),
        })
    }
 }
 /// `client → host` ([`MSG_CURSOR_RENDER`]): who renders the pointer, switched live by the
 /// client's mouse-model flip (⌃⌥⇧M — design/remote-desktop-sweep.md §8). `client_draws: true`
 /// = the DESKTOP model: the host EXCLUDES the pointer from the video and forwards
 /// shape/state ([`CursorShape`]/`0xD0`) for the client's local OS cursor. `false` = the
 /// CAPTURE model: the host COMPOSITES the pointer into the video exactly as a
 /// channel-less session would (DWM / encoder blend — full fidelity incl. XOR inversion)
 /// and the forwarder goes quiet. Sessions that negotiated the cursor cap start in
 /// `client_draws: true` (the pre-message behavior) until told otherwise; the message is
 /// idempotent and latest-wins.
 #[derive(Clone, Copy, Debug, PartialEq, Eq)]
 pub struct CursorRenderMode {
    pub client_draws: bool,
 }
 impl CursorRenderMode {
    pub fn encode(&self) -> Vec<u8> {
        // magic[0..4] type[4] client_draws[5]
        let mut b = Vec::with_capacity(6);
        b.extend_from_slice(CTL_MAGIC);
        b.push(MSG_CURSOR_RENDER);
        b.push(self.client_draws as u8);
        b
    }
    pub fn decode(b: &[u8]) -> Result<CursorRenderMode> {
        if b.len() != 6 || &b[0..4] != CTL_MAGIC || b[4] != MSG_CURSOR_RENDER {
            return Err(PunktfunkError::InvalidArg("bad CursorRenderMode"));
        }
        Ok(CursorRenderMode {
            client_draws: b[5] != 0,
        })
    }
 }
 #[cfg(test)]
 mod tests {
    use crate::config::Mode;
    use crate::quic::*;
    #[test]
    fn cursor_render_mode_roundtrip() {
        for client_draws in [true, false] {
            let m = CursorRenderMode { client_draws };
            assert_eq!(CursorRenderMode::decode(&m.encode()).unwrap(), m);
        }
        // Type byte separates it from the shape message (and vice versa).
        assert!(CursorRenderMode::decode(
            &CursorShape {
                serial: 1,
                w: 1,
                h: 1,
                hot_x: 0,
                hot_y: 0,
                rgba: vec![0; 4]
            }
            .encode()
        )
        .is_err());
    }
    #[test]
    fn reconfigure_roundtrip() {
        let rq = Reconfigure {
@@ -1147,4 +1282,42 @@ mod tests {
        assert!(ClipFetchHdr::decode(&[bytes.as_slice(), &[0]].concat()).is_err());
        assert!(ClipFetchHdr::decode(&bytes[..bytes.len() - 1]).is_err());
    }
    #[test]
    fn cursor_shape_roundtrip() {
        let s = CursorShape {
            serial: 7,
            w: 2,
            h: 3,
            hot_x: 1,
            hot_y: 2,
            rgba: (0..2 * 3 * 4).map(|i| i as u8).collect(),
        };
        assert_eq!(CursorShape::decode(&s.encode()).unwrap(), s);
        // Max-side shape still fits the u16 control frame with headroom.
        let side = CURSOR_SHAPE_MAX_SIDE;
        let big = CursorShape {
            serial: u32::MAX,
            w: side,
            h: side,
            hot_x: side - 1,
            hot_y: 0,
            rgba: vec![0xAB; side as usize * side as usize * 4],
        };
        let bytes = big.encode();
        assert!(bytes.len() <= u16::MAX as usize, "must fit a control frame");
        assert_eq!(CursorShape::decode(&bytes).unwrap(), big);
        // Rejections: zero / oversize dims, and a length that disagrees with them.
        let mut zero = s.encode();
        zero[9] = 0;
        zero[10] = 0;
        assert!(CursorShape::decode(&zero).is_err());
        let mut oversize = s.encode();
        oversize[9..11].copy_from_slice(&(CURSOR_SHAPE_MAX_SIDE + 1).to_le_bytes());
        assert!(CursorShape::decode(&oversize).is_err());
        let mut short = s.encode();
        short.pop();
        assert!(CursorShape::decode(&short).is_err());
        // Distinct from the neighboring vocabulary.
        assert!(ClipState::decode(&s.encode()).is_err());
    }
 }
--- a/Show More
+++ b/Show More