enricobuehler
4a4118e3ce
PyroWave encodes on the same GPU shader cores the game saturates, and an elevated VK_KHR_global_priority queue is the compute-preemption lever for it — measured on .21 (RTX 5070 Ti, GRID 2 loop): encode p99 6.4 -> 4.4 ms. Every driver refuses every priority class without CAP_SYS_NICE, on NVIDIA and on RADV alike, so the lever is decoration on a packaged host. 0.26.0-1 granted that capability to punktfunk-host and killed desktop streaming on every KDE box: KWin identifies a client by resolving /proc/<pid>/exe and matching an installed .desktop's Exec=, the kernel refuses that readlink to a reader whose effective set is not a superset of the target's PERMITTED set (cap_ptrace_access_check), and KWin holds no capabilities. #136 revoked it everywhere. The capability therefore cannot live in the process that fronts KWin. It lives in a new, deliberately small binary — punktfunk-encode-worker — which owns the priority-elevated Vulkan device and talks to nothing but the socket its parent spawned it on: no Wayland, no D-Bus, no network, no plugins. It is a SEPARATE FILE and must stay one; a hardlink or a hidden host subcommand shares the inode, hence the capability, and silently re-creates the incident. That rule is written where someone would break it, in the worker crate's own Cargo.toml. `open_inner` is reused verbatim in the worker — the same REALTIME->HIGH->none ladder, the same refusal-never-fails-open invariant, the same PUNKTFUNK_PERF split — so the A/B stays comparable with PW1. The only in-process change is a flag for whether THIS process prints the INERT warn, plus an out-parameter reporting the class that was granted. Three things the design did not anticipate: * An AU cannot ride in the message body. MAX_MSG is 64 KiB and bodies are serde_json, which renders a Vec<u8> as one decimal per byte: a 1080p60 AU is ~333 KB of JSON and 4K ~3.3 MB, and the minimum per-frame budget is already 64 KiB. So the AU crosses on a memfd the worker creates once and pwrites each frame; the fd crosses once, in Ready. A test pins the arithmetic so nobody "simplifies" the memfd away. Cursor bitmaps take the same route, only when their serial changes. * set_wire_chunking has to cross the wire even though poll_chunk does not. Chunking changes the AU BYTES, not merely how they are handed out — it feeds rate_budget()'s deflation and build_au's windowed framing — so a proxy-local copy would have the host cutting dense AUs at boundaries that are not window boundaries. Forwarded and mirrored. poll_chunk itself needs no protocol: the identical AuChunker runs host-side on the whole AU the worker returns. * CPU-backed frames really do reach this encoder (force_cpu_for_nvenc_444, and the raw-dmabuf degrade latch), and a 1080p BGRA frame is ~8 MB. The first non-dmabuf frame pins the session in-process with one warn rather than putting 480 MB/s on a socket. Every rung falls back to the in-process encoder exactly as today with one warn and never a dead session: PUNKTFUNK_ENCODE_WORKER=off, binary missing, spawn failure, handshake timeout, proto or workspace-version skew (host and worker are different files now, so that check is load-bearing), InitErr, a refused frame, and socket EOF mid-session — which respawns once, then pins inline. Also: recv retries EINTR with the REMAINING deadline, not a fresh one. With SO_RCVTIMEO the kernel returns EINTR rather than restarting, so a signal would otherwise read as a dead worker; re-arming with the full budget would instead let a steady signal rate defer a real hang forever.
110 lines
5.6 KiB
TOML
110 lines
5.6 KiB
TOML
# Hardware/software video encode (plan §7 / §W6): the per-vendor backends (NVENC, VAAPI, AMF, QSV,
|
|
# Vulkan-Video, PyroWave, openh264) behind one `Encoder` trait + `open_video` selector, extracted
|
|
# from the host so it depends on the shared frame vocabulary (pf-frame) rather than living inside
|
|
# the orchestrator. Speaks pf-frame (CapturedFrame/PixelFormat/dxgi identity) and pf-zerocopy
|
|
# (CUDA), never pf-capture — the capture→encode edge is one-way (plan §2.4).
|
|
[package]
|
|
name = "pf-encode"
|
|
version.workspace = true
|
|
edition = "2021"
|
|
rust-version.workspace = true
|
|
license = "MIT OR Apache-2.0"
|
|
description = "punktfunk host video encode: NVENC/VAAPI/AMF/QSV/Vulkan-Video/PyroWave/openh264 backends behind one Encoder trait."
|
|
publish = false
|
|
|
|
[dependencies]
|
|
punktfunk-core = { path = "../punktfunk-core", features = ["quic"] }
|
|
pf-frame = { path = "../pf-frame" }
|
|
pf-gpu = { path = "../pf-gpu" }
|
|
pf-host-config = { path = "../pf-host-config" }
|
|
pf-zerocopy = { path = "../pf-zerocopy" }
|
|
anyhow = "1"
|
|
tracing = "0.1"
|
|
|
|
[dev-dependencies]
|
|
# A test writer for the NVENC backend's unit tests (`with_test_writer().try_init()`).
|
|
tracing-subscriber = { version = "0.3", features = ["env-filter"] }
|
|
|
|
[target.'cfg(target_os = "windows")'.dev-dependencies]
|
|
# The QSV live e2e drives the REAL HdrP010Converter output (an RTV-written, ring-profile P010
|
|
# texture) into the encoder — the one seam the CPU-upload tests can't reach.
|
|
pf-capture = { path = "../pf-capture" }
|
|
|
|
[target.'cfg(any(target_os = "linux", target_os = "windows"))'.dependencies]
|
|
# Software H.264 (openh264, BSD-2) — the GPU-less encode path on both platforms.
|
|
openh264 = "0.9"
|
|
|
|
[target.'cfg(target_os = "linux")'.dev-dependencies]
|
|
# The encode-worker protocol tests measure what an AU costs as a serde_json body — the reason the
|
|
# access units ride a memfd instead (enc/linux/worker.rs).
|
|
serde_json = "1"
|
|
|
|
[target.'cfg(target_os = "linux")'.dependencies]
|
|
# The `punktfunk-encode-worker` protocol (enc/linux/worker.rs). The framing is pf-zerocopy's
|
|
# `ipc`, which is generic over the serde body; the message enums live here and version separately.
|
|
serde = { version = "1", features = ["derive"] }
|
|
# libavcodec (NVENC libav + VAAPI backends). `ffmpeg-sys-next` auto-detects the FFmpeg version, so
|
|
# this pin tracks the crate's own major (which shadows FFmpeg's): 9 = FFmpeg 9 (libavcodec 63,
|
|
# libavutil 61). Arch shipped FFmpeg 9 on 2026-08-08 and every soname moved with it; the packaged
|
|
# host must be BUILT against the FFmpeg it will run on, and packaging/arch/PKGBUILD now derives a
|
|
# soname dep from that link so pacman can no longer walk an install across the break.
|
|
ffmpeg-next = "9"
|
|
libc = "0.2"
|
|
# Vulkan bindings for the raw Vulkan-Video encode + PyroWave compute backends (feature-gated below;
|
|
# the dep stays unconditional to mirror the host's Linux target — unused-but-declared is harmless).
|
|
ash = "0.38"
|
|
# `libnvidia-encode.so.1` is dlopen'd at runtime for the direct-SDK NVENC/CUDA backend.
|
|
libloading = "0.8"
|
|
# Direct-SDK NVENC (raw `sys::nvEncodeAPI` types; entry points resolved at runtime). `ci-check` =
|
|
# vendored bindings, no CUDA toolkit at build.
|
|
nvidia-video-codec-sdk = { version = "0.4", features = ["ci-check"], optional = true }
|
|
# PyroWave (opt-in wired-LAN wavelet codec) — vendored codec + bindgen'd C API, only under `pyrowave`.
|
|
pyrowave-sys = { path = "../pyrowave-sys", optional = true }
|
|
|
|
[target.'cfg(target_os = "windows")'.dependencies]
|
|
# NVENC (direct SDK, D3D11 input) + the shared D3D11/DXGI vocabulary via pf-frame.
|
|
nvidia-video-codec-sdk = { version = "0.4", features = ["ci-check"], optional = true }
|
|
# AMD (AMF) + Intel (QSV) hardware encode via libavcodec (behind `amf-qsv`; link-imports FFmpeg).
|
|
ffmpeg-next = { version = "9", optional = true }
|
|
# `libnvidia-encode`/`nvEncodeAPI64.dll` resolved at runtime; the NVENC status→cause table dlopen.
|
|
libloading = "0.8"
|
|
# Native Intel QSV (VPL): vendored static MIT dispatcher + bindgen'd C API, only under `qsv`.
|
|
libvpl-sys = { path = "../libvpl-sys", optional = true }
|
|
# PyroWave (opt-in wired-LAN wavelet codec) — vendored codec + bindgen'd C API, only under
|
|
# `pyrowave`. The Windows backend is the NV12 zero-copy D3D11→Vulkan encoder; same crate as Linux.
|
|
pyrowave-sys = { path = "../pyrowave-sys", optional = true }
|
|
windows = { version = "0.62", features = [
|
|
"Win32_Foundation",
|
|
"Win32_Graphics_Direct3D",
|
|
"Win32_Graphics_Direct3D11",
|
|
"Win32_Graphics_Dxgi",
|
|
"Win32_Graphics_Dxgi_Common",
|
|
# SECURITY_ATTRIBUTES — the PyroWave backend's IDXGIResource1::CreateSharedHandle signature.
|
|
"Win32_Security",
|
|
"Win32_Storage_FileSystem",
|
|
"Win32_System_LibraryLoader",
|
|
"Win32_System_Threading",
|
|
# D3DKMTSetProcessSchedulingPriorityClass — raise the host's WDDM GPU scheduling priority
|
|
# above a running game so PyroWave's compute-shader encode isn't starved (enc/windows/pyrowave.rs).
|
|
"Wdk_Graphics_Direct3D",
|
|
] }
|
|
|
|
[features]
|
|
default = []
|
|
# NVENC hardware encode (Linux CUDA + Windows D3D11); entry points resolved at runtime.
|
|
nvenc = ["dep:nvidia-video-codec-sdk"]
|
|
# AMD (AMF) + Intel (QSV) hardware encode on Windows via libavcodec.
|
|
amf-qsv = ["dep:ffmpeg-next"]
|
|
# Raw Vulkan-Video HEVC/AV1 encode on Linux (reuses the `ash` bindings; no new dep).
|
|
vulkan-encode = []
|
|
# PyroWave — the opt-in wired-LAN intra-only wavelet codec (Linux encode backend).
|
|
pyrowave = ["dep:pyrowave-sys"]
|
|
# Native Intel QSV via the statically linked VPL dispatcher (Windows; runtime GPU
|
|
# libs come from the Intel driver store). Supersedes the ffmpeg `amf-qsv` QSV path
|
|
# (design/native-qsv-encoder.md). ⚠ Like `nvenc`: hand builds need this feature or
|
|
# Intel boxes fall through to the ffmpeg path / software.
|
|
qsv = ["dep:libvpl-sys"]
|
|
|
|
[lints]
|
|
workspace = true
|