Y2E-140 — Establish a reliable first-boot observation path

Depends on Y2E-125/130 and the completed M1 offline artifact. Research directly; supporting host tooling only. No candidate flash, pin probing, voltage injection, boot-mode change or feature expansion.

Acceptance:

• Reconcile package preloader/LK/kernel UART selection with safe installed-stock console evidence; record exact normal/conditional selection and baud/format.
• Establish a board-specific physical TX/GND connection and logic voltage from credible evidence, or explicitly retain this blocker and evaluate the smallest alternative.
• Define reproducible receive-only host capture with retained raw bytes, timing, adapter/settings and integrity metadata.
• Capture identifiable stock LK/kernel output before changing BOOTIMG; distinguish a host tooling test from device proof.
• Decide whether the existing artifact needs revision; any justified change must rerun all offline validation.
• Update authoritative knowledge and identify the next bounded launch proof. Keep this issue open if physical/capture acceptance remains unsatisfied.

2026-09-08 result: OPEN, physical evidence required. Installed ttyMT0/sysfs/iomem plus remapped FM driver prove the normal installed-kernel controller is UART0 (0x11002000), superseding UART3. Package LK's conditional UART0 branch is consistent with this result, but installed handoff identity/routing and actual baud are not fully proved. The manufacturer native pin-domain table supports expected 1.8 V; no board-specific TX/GND pad or pad measurement is confirmed. No adapter was present.

D10 corrects only DT UART0/IRQ51 and strict validation. Two fresh full offline builds match across 11 outputs; all nine artifact tests pass (22 real-artifact and 25 package mutation/extent cases), source verification covers 91,166 files/links. All D08 addresses/sizes, kernel/config/initramfs bytes stay unchanged. Current BOOTIMG SHA-256: 574061649e10266927267b68afbc95da7fe0761dc9823e88767741bac9363fd9.

Receive-only host tool/runbook implemented; four PTY tests pass on Python 3.14.7 and locked 3.12.14. These are not hardware proof. Actual ADB /proc/last_kmsg yields only a prior late kernel/shutdown log, not LK/serial capture. Next missing evidence is verified UART0 TX/GND/level plus actual stock LK/kernel capture. If native TX is inaccessible, investigate exact preloader UART-over-USB accessory detection/electrical/routing persistence; do not guess cables or switch PHY live. U07b is the next launch gate after observation passes; U07c and backup/recovery/power remain required.

Authoritative evidence, capture runbook, revised artifact. No candidate boot/flash or additional Linux feature occurred.