docs: mark session-aware follow-ups #2 (switch input) + #3 (vout primary) resolved
android / android (push) Failing after 21s
ci / web (push) Failing after 3s
ci / docs-site (push) Failing after 1s
ci / bench (push) Failing after 0s
deb / build-publish (push) Failing after 0s
decky / build-publish (push) Failing after 1s
docker / build-push (--build-arg FEDORA_VERSION=44, ci, ci/fedora-rpm.Dockerfile, punktfunk-fedora44-rpm) (push) Failing after 0s
docker / build-push (., web/Dockerfile, punktfunk-web) (push) Failing after 1s
docker / build-push (ci, ci/fedora-rpm.Dockerfile, punktfunk-fedora-rpm) (push) Failing after 0s
docker / build-push (ci, ci/rust-ci.Dockerfile, punktfunk-rust-ci) (push) Failing after 1s
docker / build-push (docs-site, docs-site/Dockerfile, punktfunk-docs) (push) Failing after 1s
apple / swift (push) Successful in 53s
ci / rust (push) Failing after 2m33s
docker / deploy-docs (push) Has been skipped
flatpak / build-publish (push) Failing after 0s
rpm / build-publish (bazzite, punktfunk-fedora-rpm) (push) Failing after 1s
rpm / build-publish (fedora-44, punktfunk-fedora44-rpm) (push) Failing after 0s

Both landed in 3363576 and validated live on the Bazzite F44 box: a Gaming→Desktop
mid-stream switch shows `settled desktop portal env … compositor=kwin` →
`portal granted devices` → `device RESUMED` (input lands, no reconnect), and
`KWin: streamed output set as the sole desktop also_disabled=["HDMI-A-1"]` (panels
on the streamed screen). Remaining: #1 (F44 gamescope teardown GPU leak) + the
lower-priority polish.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-06-15 07:12:29 +00:00
parent 336357643c
commit 38cce754bd
+14 -23
View File
@@ -8,10 +8,20 @@ backend in place without a reconnect.
Live-validated on the Bazzite F44 box (`bazzite-deck-nvidia:testing`, RTX 4090): Desktop KDE at Live-validated on the Bazzite F44 box (`bazzite-deck-nvidia:testing`, RTX 4090): Desktop KDE at
5120×1440 + input; Gaming managed at 5120×1440; warm-session reuse on quick reconnect; Feature B 5120×1440 + input; Gaming managed at 5120×1440; warm-session reuse on quick reconnect; Feature B
video-switch both directions. The items below are **deliberately parked** — they have workarounds video-switch both directions.
and/or are F44-specific.
## High priority ## Resolved (2026-06-15, `3363576`)
- **#2 — mid-stream-switch input** ✅ `vdisplay::settle_desktop_portal()` pushes the live session env
into the systemd/D-Bus activation environment and restarts the KWin portal on a switch, so input
lands without a reconnect. Validated live: `settled desktop portal env … compositor=kwin`
`libei: portal granted devices``device RESUMED` on a Gaming→Desktop mid-stream switch.
- **#3 — KWin/Mutter virtual output primary** ✅ `apply_session_env` defaults
`PUNKTFUNK_KWIN_VIRTUAL_PRIMARY` / `PUNKTFUNK_MUTTER_VIRTUAL_PRIMARY` on for the auto desktop path.
Validated live: `KWin: streamed output set as the sole desktop also_disabled=["HDMI-A-1"]` — panels
now render on the streamed screen.
## Still parked
### 1. F44 gamescope teardown corrupts the GPU context ### 1. F44 gamescope teardown corrupts the GPU context
Every gamescope teardown on this box (stop the autologin on connect; stop the managed session on Every gamescope teardown on this box (stop the autologin on connect; stop the managed session on
@@ -31,26 +41,7 @@ but don't eliminate it. Options, in order of preference:
connected (the autologin is never restored; return to gaming manually). connected (the autologin is never restored; return to gaming manually).
- Upstream gamescope/driver fix. - Upstream gamescope/driver fix.
### 2. Mid-stream-switch input (Gaming→Desktop) is flaky until a reconnect (#2 mid-stream-switch input and #3 virtual-output-primary are **resolved** — see the Resolved section above.)
After a mid-stream switch to a desktop, pointer/keyboard often don't land until a disconnect+reconnect.
Root cause: on the switch the host opens the KDE `RemoteDesktop` **portal** immediately, but the
portal is a D-Bus-activated service that reads the **systemd `--user` activation environment**, which
hasn't settled to the new session yet — so the portal session is created against a half-stale env. It
*accepts* events (no error, so the injector's reopen-on-failure never fires) but they don't reach
KDE. The host log even shows `libei: portal granted devices` + `device RESUMED`, yet input is dead.
**Workaround: reconnect once after switching** (re-opens the portal against the settled env → works).
Fix: on a session switch, push the new session env into the systemd/D-Bus activation environment
before reopening input — `systemctl --user import-environment WAYLAND_DISPLAY XDG_CURRENT_DESKTOP
DBUS_SESSION_BUS_ADDRESS` + `dbus-update-activation-environment` (exactly what a fresh KDE login does;
see `scripts/headless/run-headless-kde.sh:118-119`) — and/or delay/retry the input reopen until the
portal is settled.
### 3. KWin virtual output is not set as primary
The KWin virtual output isn't marked the primary display, so KDE panels / primary-screen content can
stay on the (now-absent) original output instead of the streamed virtual screen. Follow-up: set the
virtual output primary on create — `vdisplay/kwin.rs` already reads `PUNKTFUNK_KWIN_VIRTUAL_PRIMARY`;
make it default-on (or always promote the new output to primary via the KWin/kscreen API) so the
streamed screen is the primary desktop.
## Lower priority / polish ## Lower priority / polish