enricobuehler f9b857aac2 feat(capture): true SHM path (PUNKTFUNK_FORCE_SHM) for race-free Mutter+NVIDIA ...

Empirically, Mutter+NVIDIA dmabuf capture has NO working GPU sync — confirmed on
worker-3: explicit sync fails buffer alloc (EINVAL, no cogl sync_fd), and the
dmabuf carries no implicit fence (EXPORT_SYNC_FILE waited=false). So any dmabuf
read — zero-copy import OR mmap — races Mutter's render and flashes the buffer's
previous frame. The prior "CPU fallback" still listed DmaBuf in its buffer types,
so Mutter kept handing dmabufs and it never fixed anything (got worse).

PUNKTFUNK_FORCE_SHM=1 offers MemPtr+MemFd ONLY (no DmaBuf), forcing Mutter to
glReadPixels-download into mappable memory — which orders against its render, so
the frame is complete + current by construction (race-free). Costs the download
(~3 ms) + zero-copy; correct at 1080p/4K60. KWin/gamescope are unaffected (they
blit into the buffer, no read-before-render race) and keep zero-copy.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

2026-06-13 09:35:28 +00:00

..

punktfunk-client-linux

feat(client-linux): VAAPI hardware decode — zero-copy dmabuf into GraphicsOffload

2026-06-12 23:26:59 +00:00

punktfunk-client-rs

fix(client-linux): absolute mouse was dropped — pack the surface size in flags

2026-06-12 20:50:53 +00:00

punktfunk-core

perf(core): recvmsg_x batched receive on Apple (macOS client)

2026-06-12 23:52:39 +00:00

punktfunk-host

feat(capture): true SHM path (PUNKTFUNK_FORCE_SHM) for race-free Mutter+NVIDIA

2026-06-13 09:35:28 +00:00