evidence · 3.3 KiB

README.md

Committed at docs/evidence/mainline-display/physical-2026-10-01/corrected-probe-recovery-2026-10-02/README.md. This is what grounds the requirements that cite it.

Open raw source

Recovery after the corrected initrd diagnostic

The operator reported “i power cycled” after the bounded 2026-10-02 mainline rdinit=/bin/sh trial reached Linux but returned no complete fresh-token probe result or observed reboot. The coordinator then reserved the board and serial port for an independent normal-system recovery check. Raw serial captures remain protected outside Git; no raw credentials or addresses are published.

At 2026-10-02T06:44:06.065931+00:00, the unchanged fail-closed checker from source revision b209129b1cf681669edfe3e1e05be317842bf6f4 passed on the physical board. postflight.json records the exact p1a1hz… normal system and persistent profile, selected 03zyl0…/Image, Linux 6.6.36, one matching init= selector, three active shell services, all eight protected boot-file hashes and a new boot ID e7cf8096-44a0-4bdf-815d-2f1aec4126fb. The previous independently observed normal ID was f3940c26-ce09-41f5-b53f-fe525a04db06; the checker required a different ID. The expected system/profile/kernel and hashes came from the committed ordinary-boot baseline, with that previous normal ID supplied from the first recovery.

The helper and expected data were streamed only into /run over an exclusive 115200-baud /dev/ttyACM0 session holding /tmp/k230-board.lock. The exact helper SHA256 is 86bcf6f2c175d44d27f68015a9cc5a1ad7731ad50b6bbf3347859a4840cc7df7. Command:

/nix/store/v189xydz6qkcd4cbkixcmv91w8hbc560-python3-riscv64-unknown-linux-gnu-3.14.7/bin/python3 -I /run/k230-mainline-normal-state.py postflight FRESH_32_HEX_TOKEN

Only a complete standalone fresh-token identity marker was accepted. No normal boot-file, persistent profile or stage-1 mutation was performed by this recovery check. The selected normal boot configuration survives the power cycle; this is not an installed mainline system.

The coordinator requested compositor output * dpms on and card_shell home, then used the exact installed Rust shell executable 3hy6h165… with --surface hide to hide transient surfaces. output-state.json records DSI-1 active/powered, 568×1232 at 52.190 Hz, transform normal. home.png is a reviewed full native grim capture from the running Wayland session. home-panel.jpg is a reviewed camera observation of the illuminated physical panel; the camera frame covers only part of the panel, while the native image shows the full frame. Camera command:

ffmpeg -nostdin -hide_banner -loglevel error -f v4l2 -input_format mjpeg -video_size 1280x720 -i /dev/video0 -vf 'crop=600:530:680:190,transpose=2,transpose=2' -frames:v 1 -q:v 2 -y "$HOME/tmp/k230-reset-recovery/second-home-panel.jpg"

Home was requested through IPC. These are physical recovery, serial identity, native capture and camera evidence, not a new real-finger test. The prior normal-boot navigation acceptance remains separately recorded in operator-navigation.json. This recovery does not identify why the candidate probe stopped, establish partition/IRQ state, or satisfy mainline root-login/touch task 5b.5.