23 Commits
Author SHA1 Message Date
Jeff Zhang 2dd3280b7e perf(inproc): replace portal pseudo-polling with event-driven frame wait (#193)
* perf(portal): wait for frames without polling

* fix(portal): use monotonic clock for condition variable wait deadline

* fix(portal): compute wait_for_first_frame deadline with monotonic clock
2026-11-06 10:41:21 +08:00
Jeff Zhang b0015a6c48 perf(input): offload text_input to spawn_blocking preserving order and rebase on main (#189) 2026-11-06 05:39:20 +08:00
Jeff Zhang c60d9f1bfa perf: HEVC RTP 分包零拷贝子切片与 X11 终端检测 100ms 缓存 (#182)
* perf: HEVC RTP 分包零拷贝子切片与 X11 终端检测 100ms 缓存

* fix(x11): 绑定焦点窗口缓存并支持焦点切换即时失效
2026-11-06 04:38:00 +08:00
Jeff Zhang 6032129228 perf: 优化 H.265 IRAP 单趟流式扫描与 CONTROL 通道 JSON 单次解析 (#180) 2026-11-06 03:38:47 +08:00
Jeff Zhang 7ba71a5540 webrtc: single-pass decode_wire for data channel packets and merge loop passes (#191) 2026-11-06 03:38:37 +08:00
Jeff Zhang b774bf6b16 perf: CompleteFrame 使用 bytes::Bytes 消除每帧码流克隆深拷贝 (#184) 2026-11-06 03:38:26 +08:00
Jeff Zhangandferstar b9586f226d input: park the pointer worker until pointer or gesture work exists (#172)
* input: park the pointer worker until pointer or gesture work exists

A fixed 4 ms tick kept ~250 wakeups/s (clock_gettime storms, worker
steal spins, pow-heavy flush math) burning ~3% CPU on an idle host with
no viewer connected. The worker now parks on a Notify permit until a
pointer/scroll/button event or a live touch gesture needs the 4 ms
cadence; notify_one permit semantics make the park check race-free.
Sub-notch scroll residue cannot inject alone, so it parks too.

* fix(input): preserve pointer wake ordering and motion cadence

Hold the serial lock before notifying button transitions so a parked worker cannot miss a newly pending gesture. Keep the 4 ms coalescing cadence until an empty tick and reset the interval after each parked phase.

Add regression coverage for button wake ordering, 1 ms motion streams, and idle cadence reset using Tokio test-only clock control.

---------

Co-authored-by: ferstar <[email protected]>
2026-11-04 23:18:19 +08:00
Jeff Zhang fcfd2b6717 feat(wallpaper): discover Omarchy backgrounds automatically (#157)
Read Omarchy's current/background link (XDG_STATE_HOME, falling back to XDG_CONFIG_HOME) on Hyprland; explicit UU_WALLPAPER keeps precedence.
2026-11-03 12:31:33 +08:00
Jeff ZhangandJeff Zhang 3bce605eaa feat(host): 上报系统壁纸,补齐设备卡片缩略图 (#153)
* feat(host): report system wallpaper for device cards

* fix(host): handle quoted wallpaper paths and KDE defaults

* docs: align wallpaper guidance and disclose default uploads

* fix(host): validate active KDE plugin for stored wallpapers

* fix(host): support Plasma 5 wallpaper config queries

* fix(host): exclude Plasma panels from wallpaper reporting

* fix(host): follow image symlinks in KDE wallpaper packages

---------

Co-authored-by: Jeff Zhang <[email protected]>
2026-11-03 01:25:57 +08:00
Jeff Zhang f4f16aafe4 fix(privacy): synchronize Mac viewer privacy status over TEXT (#148)
* fix(privacy): synchronize Mac viewer privacy status over TEXT

* fix(webrtc): bound pending TEXT replies and coalesce metrics
2026-11-02 11:17:55 +08:00
Jeff Zhang 8e30371317 feat(host): KDE Wayland privacy screen (RPC 7) (#146)
* feat(host): KDE Wayland 被控端防窥模式(隐私 / 关屏)

viewer RPC 7 打开防窥时,采集和输入切到 KWin 虚拟输出(portal ScreenCast VIRTUAL),
再用 kscreen-doctor 禁用物理屏;隐私模式随后重新启用物理屏并全屏覆盖桌面壁纸。
防窥期间通过 KWin InputDevice 禁用本地键鼠,指针/按键/文字改走 uinput
(KDE 不处理虚拟输出共享上的 portal 输入)。

- privacy_type 实测:1 关闭、2 隐私(壁纸)、3 关屏;SettingMetric 4/7 上报真实状态
- 恢复记录 + privacy-guard 子进程:主机以任何方式结束(含 SIGKILL)都恢复屏幕和输入
- src/privacy_screen/ 分层:控制器/协议/记录/守护/遮挡窗口共享,桌面实现走 Backend trait(kde/)
- portal:Session::set_virtual,闭包恢复按当前源重开;编码器随 portal 尺寸变化重建

Refs #140

* fix(privacy): 按 Mac 客户端协议应答防窥请求,修复授权竞态

- FeatureFlag ff_private_screen=2(ENHANCED_PRIVACY),声明支持关屏
- 先回 RPC 7 应答再切换,避免客户端超时报 9998;切换失败发 ReportError -7
- 应答 RPC 25 update_screen_saver_req(RpcResponse 21),隐私模式不再被中止
- SettingMetric 4/7 上报开关与最近一次模式(host_settings.json privacy_type),状态变化推送 SystemStateChange 5
- portal 会话串行打开,打开失败回填一次性授权:快速重连时防窥切换与闭包恢复并发曾清空保存的授权

字段号与枚举取自官方 Mac 客户端的 SwiftProtobuf name map。
已知限制:Mac 客户端防窥菜单切换后仍显示"不启用",实际模式正确。

Refs #140

* fix(privacy): 处理 #146 评审意见

- 预检 uinput:injector 未打开 uinput(/dev/uinput 不可用或 UU_INPUT_BACKEND=native)时,
  在改动输出和设备之前拒绝防窥(虚拟输出共享上 KDE 忽略 portal 输入)
- leave() 中输入设备恢复失败在恢复输出后向上返回,保留 privacy-power.json 与 guard
- record::ownership() 只缓存成功持锁,guard 轮询和 recover() 可重试
- 遮挡窗口就绪握手:privacy-wallpaper 建好窗口后输出 ready,5 秒内未就绪则报错并回滚
- rebase 到 main(#145):DMA-BUF 打开路径跳过虚拟输出、失败同样回填授权;
  #145 的 uinput 指针路径也按隐私模式的虚拟输出区域换算坐标
- README 表格压缩为一句,细节在 docs/privacy-screen.md

Refs #140

* docs(privacy): 注明 ff_system_metrics=3 对应的枚举

Refs #140

* fix(privacy): resolve lifecycle and restoration review findings
2026-11-02 01:14:10 +08:00
Jeff Zhang 73293318b4 feat(portal): VAAPI DMA-BUF zero-copy, in-process Opus audio and lock-screen uinput control (#145)
* feat(audio): capture and encode Opus in process

* feat(portal): encode VAAPI from leased DMA-BUF frames

* fix(input): route Portal controls and committed ASCII through uinput

* docs(portal): shorten setup notes and colocate hardware acceptance guidance

* fix(input): preserve GNOME IME guard for committed uinput text

* build: link pulse/opus after the static C archive (--as-needed)
2026-11-01 22:12:04 +08:00
Jeff Zhangandferstar 33861082dc build: link inproc-av dependencies after the static engine archive (#105)
GNU ld with --as-needed (Ubuntu default) drops shared libraries listed before
libx11_engine.a, so --features inproc-av failed to link with undefined gbm_*,
xcb_dri3_* and xcb_damage_* on Ubuntu (found on aarch64; Arch is unaffected).

Co-authored-by: ferstar <[email protected]>
2026-11-01 20:11:39 +08:00
Jeff Zhangandferstar 8792fb4eca vaapi: raise the in-process RC max_qp to 51 so capped presets reach the QP 51 floor (#102)
Co-authored-by: ferstar <[email protected]>
2026-10-23 11:24:15 -07:00
Jeff Zhangandferstar 6a3da14961 Switch the desktop encoder on SIGUSR2 without dropping the session (#96)
set-encoder / set-codec signal the running host (only when /proc/<pid>/exe is this exact binary); the live desktop stream re-plans its encoder at the next frame boundary, sends CaptureSettingSync when 4:4:4 falls back, and prints a note when no host was notified. Step 3 of #85.

Co-authored-by: ferstar <[email protected]>
2026-10-23 03:24:29 -07:00
Jeff Zhang 9d3b3da3f5 Video: probe the auto encoder in the background (NVENC then CPU) (#93)
Refs #85 (step 2).
2026-10-23 00:24:31 -07:00
963875516a Video: encoder and codec selection settings (set-encoder / set-codec) (#92)
Step 1 of #85: unified encoder/codec selection (host_settings.json > UU_VIDEO_* > auto), set-encoder / set-codec subcommands with locked merge writes, per-connection selection snapshot, doctor shows sources.

Co-authored-by: ferstar <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-22 16:24:22 -07:00
Jeff Zhang 9e2198b03d docs: AMD VAAPI verified on the in-process engine; CPU cost figures (#81)
README/README.en/AGENTS: AMD iGPUs (HawkPoint, Cezanne) are accepted on the inproc-av VAAPI engine with the official
Mac viewer; the FFmpeg process pipeline on AMD is not separately verified. inproc-av now covers VAAPI in process;
build headers listed. Thanks @ferstar.
2026-10-21 22:24:42 -07:00
Jeff Zhang 21c01d2e18 inproc-av: native VAAPI H.264/HEVC through DRI3 DMA-BUF, VPP and VCN (#80)
With --features inproc-av on X11 and UU_VIDEO_ENCODER=vaapi, VAAPI runs in process: a DRI3 DMA-BUF of the root
(XShm upload without DRI3) goes through VPP to NV12 and the hardware encoder. SPS/PPS(/VPS) and slice headers are
packed by the host because radeonsi (Mesa 26) does not write them. libva/libva-drm are dlopen'ed; release builds do
not enable inproc-av. The NVENC path is unchanged.

Accepted by the contributor on AMD HawkPoint1 (Mesa 26.2.3), 1920x1200@60, with the official Mac viewer
(HEVC, presets, fps, scaling). Intel iHD and 4K are not verified. Thanks @ferstar.
2026-10-21 15:25:13 -07:00
059a3b1bb8 inproc-av: 4:4:4 on the native NVENC engine (#76, #77)
H.264 High 4:4:4 / HEVC FRExt with chromaFormatIDC 3 and AQ off when 真彩 is requested; GPUs without YUV444 encode fail the capability check and fall back. full_chroma_available() no longer excludes the in-process path. UU_PROBE_CHROMA=444 for the probe.

Co-authored-by: ferstar <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-21 08:24:32 -07:00
76941459cc inproc-av: native in-process NVENC reading XShm directly (#74, #75)
Replaces the FFmpeg-based in-process encoder with a native NVENC engine (vendored nv-codec-headers n12.1.14.0, libcuda/libnvidia-encode dlopen'ed at runtime, primary context reuse). Fallback: native NVENC -> FFmpeg NVENC process -> libx264. inproc-av no longer links FFmpeg; VAAPI/CPU always use the process pipeline. Default builds are unchanged.

Co-authored-by: ferstar <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-21 07:26:45 -07:00
25893cd4ab docs: align the user systemd unit with the deb package unit (#69)
Drop the hard-coded DISPLAY=:0 and install under graphical-session.target like packaging/deb; Xfce/i3, headless and the virtual desktop keep WantedBy=default.target. AGENTS.md recipe aligned.

Co-authored-by: ferstar <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-20 22:42:11 +08:00
Jeff Zhangandferstar 169edf2224 inproc: pad odd native captures to even for the in-process encoder (#60)
The in-process encoder opened its context at the raw capture size, so an
odd logical desktop (e.g. two stacked 150% monitors, 2560x2507) made every
H.264/HEVC/NVENC encoder fail to open. The encoder size is now rounded up to
even, and a native capture is copied into a zeroed frame of that size so the
extra row/column stays black. Scaled output paths are unchanged.

Co-authored-by: ferstar <[email protected]>
2026-10-20 07:23:54 +08:00