Fixed initrd debug logging: physical comparison

Physical UART evidence at revision d1cd3ca0e5c6227e71e465bc116d1e8082bca689, captured 2026-10-05 02:26:45–02:30:11 UTC. Total invocation 205.575 seconds includes preparation/loading; passive readiness was bounded to 180 seconds. Exit 1 records ordinary init login readiness unverified.

Reviewed actual host qualification passed with unchanged p2/24h artifacts, selected source/config/Image/DT/archive/manifest, matching systemd init and five load/CRC expectations. The preceding NEW protected ordinary recovery passed exact identities, distinct boot, eight hashes, three services and registration absence. The begin invocation performed its own protected normal preflight before reboot/loading.

One board/UART reservation ran:

python3 tools/mainline-drm-system-trial.py begin \
  --wait-initramfs-in-initcall --without-boot-markers --initrd-debug-logging \
  --bundle /nix/store/p2kdar89q7dhwdajjy0mrhmxxzcgsalw-k230-mainline-drm-trial-boot-files \
  --manifest PRIVATE_MANIFEST --normal-report NEW_PRIVATE_NORMAL \
  --state NEW_PRIVATE_STATE --log NEW_PRIVATE_LOG --result NEW_PRIVATE_RESULT

Fresh Linux and Run /init as init process appeared. That kernel announcement precedes execution success; it is not proof systemd ran or a runtime PID1 guard. No systemd startup/version, closure-unit or login messages were observed within the bound. This comparison stopped observably earlier than the quiet ordinary run, which showed mounted sysroot and starting closure lookup. Additional logging can change timing or console pressure. Absence of a line identifies neither a blocked instruction nor a logging/backend/hardware cause.

Result preserves safe facts and private-log hashes. Fixed observations are derived, not a raw transcript. No camera or real-glass proof was obtained for this comparison. Independent full-transcript review PASS: five strict load counts/CRCs, exact printed and fresh received arguments, source hash, unique protected preflight and replayed readiness failure match. Command annotations and source show no input after bootm; this is not independent electrical TX-direction proof.

All reservations are released. A subsequent NEW operator reset passed protected normal recovery: distinct boot, exact identities, eight matching hashes, three active services and registration absence. No previous reset was reused, fallback input or automatic return claimed. Ordinary root, panel/glass and task 5b.5 stay UNVERIFIED. The next bounded plan considers a one-value debug→info comparison with the same console destination; it is not run.