evidence · 2.5 KiB

source-preflight.md

Committed at docs/evidence/hdmi-hotplug/live-switch/source-preflight.md. This is what grounds the requirements that cite it.

Open raw source

No-reboot switching: source preflight

Recorded 2026-10-09T14:53:11.176690+00:00. Evidence class: local kernel source inspection only. No device tree was trial-booted and no physical plug/unplug transition was observed at this inspection. The accepted HDMI trial is prior evidence; the subsequent board preflight found the protected panel boot running (see panel-raw-hpd.json).

The existing LT9611 driver cannot be made a standalone monitor merely by adding its I2C node to the panel device tree:

Task 4.1 therefore needs a monitoring path that avoids attaching a second DSI consumer and leaves touch’s shared reset/IRQ ownership intact. Its physical proof remains open: the monitor must probe alongside the active panel, and actual cable changes must be checked. The HPD register’s behavior in that mode is still UNVERIFIED. Do not enable the shared HDMI IRQ merely because its electrical drive type remains unknown.

This does not establish that seamless switching is impossible. After a successful task-4.1 board proof, task 4.2 requires explicit kernel design for switching the DSI consumer and compositor-visible output. The optional reboot-based Settings prototype is not a substitute for that result.

Source artifacts at inspection:

Inspection command:

rg -n 'parse_dt|probe|detect|attach_dsi' nix/patches/mainline/drm/lontium-lt9611-k230.c
rg -n 'find_panel|bridge|attach|detach|bind' nix/patches/mainline/drm/canaan_dsi.c

Task 4.1 is not complete; no hardware outcome is inferred from this reading.