Skip to content

device-oriole: add minimal first-light policy package - #222

Merged
derekross merged 8 commits into
derekross:mainfrom
Gr-linkk:oriole-first-light-package
Nov 11, 2026
Merged

derekross merged 8 commits into
derekross:mainfrom
Gr-linkk:oriole-first-light-package

Conversation

@Gr-linkk

@Gr-linkk Gr-linkk commented Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

Offline follow-up to the 2026-11-08 review

This package alone is not yet a supported Pixel 6 install. README and device.conf now state that an image built from this package omits the private startup/recovery safeguards. This takes the documentation option offered in the review; no private services, battery code, boot-success writes or watchdog configuration are added.

The audited private boot check validates staged storage, compositor, shell and touch state, then disarms its bounded handoff watchdog and clears a persistent Linux hang marker. The matching initramfs uses a leftover marker to enter guarded USB rescue before providers. It does not mark B successful or reset ABL's retry budget. The README explains the unmarked-slot limit and why stock-A fallback cannot be assumed to recover a phone whose userdata now contains Linux.

The private check depends on unpublished initramfs assets and a matching kernel. RebootWatchdogSec=1min is a separate shutdown safeguard; the recovered watchdog prerequisites do not establish standalone public-kernel binding/shutdown-reset qualification. It remains deferred rather than importing private integration code or claiming unsupported recovery coverage.

The optional refusal test now passes relative-dir instead of /, so a broken destination guard would not target the build host's root. Package contents/version and installed runtime behavior are unchanged.

Validation: package test/all, repository tests/all, repository policy and whitespace checks pass. No phone access or new hardware qualification occurred. Derek accepted the unsupported-standalone scope on 2026-11-09 and confirmed this PR is ready for review. Standalone boot/recovery and package-combination boot/live rollback remain unqualified; they are limitations of this minimal package, not remaining readiness gates for this PR. The historical installation/smoke evidence below used a private companion and does not establish a supported standalone install.


Adds a minimal Pixel 6 first-light policy package: reboot/system-setup helpers, storage-guard continuation, USB fallback networking, disabled sleep modes and six coldplug refusals. No services are enabled. Includes keyring dependency, device provides/conflicts, repository mirrorlist and channel-preserving install hook; omamobile-bootctl is omitted.

Version 0.2.0 sorts above the installed local 0.1.9-2 and satisfies the hardware packages' existing minimum-version dependencies. Native makepkg build/check/package passed with a separate BUILDDIR, and local package tests pass.

On 2026-11-07, normal signed pacman -U installed 0.2.0-1 on the Pixel 6, together with a PRIVATE companion preserving the existing boot-check/battery scripts and units, power-key policy and reboot-watchdog policy. This companion is not included in this PR; the public package must not replace a local integration package alone. All package signatures were verified by the phone's existing keyring; dependencies, signatures and hooks stayed enabled. Offline pacman replacement/rollback ownership tests pass.

Fresh SSH, safeguard hashes/ownership, unchanged enabled units, repository configuration/channel and pacman database checks passed. The supported systemd update marker temporarily prevented a udev retrigger. Laptop logging and a bounded rollback timer covered installation; owned safety markers/timer were cleared after checks. No reboot, flash, provider retry or shell restart.

Historical smoke has existing error-level kernel messages and backlight/NFC startup refusals; this is not a clean standalone-image or boot qualification. Live rollback and a boot of the new package combination remain untested. See device/oriole/NOTES.md's 2026-11-07 installation entry for the safeguards and evidence.

Post-install phone-smoke report (integration system with private safeguard companion)

phone-smoke: oriole

Kernel 7.3.0-rc6-oriole-hsi1-late-dev0-g4bd35d34abb9-dirty, tools at pr222-0.2.0, 2026-11-07T17:46Z.

Check Result Detail
boot (required) ok kernel 7.3.0-rc6-oriole-hsi1-late-dev0-g4bd35d34abb9-dirty, slot b, up 6214s
storage guard (required) ok 49 partitions; only userdata and slot B's boot images writable
root (required) ok ext4 on userdata, 97G free
kernel health (required) ok no oops or BUG; 0 WARN, 38 error lines in dmesg (19 known, 19 new)
new dmesg errors not ok maxtcpc N: Failed to get vbus regulator handle x8; OF: overlay: WARNING: memory leak will occur if overlay removed, property: /soc@N/usi@N/status x2; OF: overlay: WARNING: memory leak will occur if overlay removed, property: /soc@N/usi@N/N@N/status x2
systemd units not ok failed: omamobile-oriole-backlight.service omamobile-oriole-nfc.service
USB gadget (required) ok bound to 11110000.usb
display ok card0-Unknown-1 connected, 1080x2400
GPU ok render node on panfrost
session (required) ok Hyprland and the Omarchy shell running
lock ok PAM file present (the shell didn't answer lock status)
touch ok ST FTM5 touchscreen (oriole staged)
buttons ok gpio-keys
Wi-Fi ok wlP1p1s0 on brcmfmac, link up
Bluetooth ok hci0, powered true
battery ok 70% Charging, 30 °C
sound cards skip none
sensors skip no sensor proxy
cameras skip no libcamera tools
thermal ok hottest zone 30 °C

Packages: hyprland 0.56.2-7, libcamera 0.7.2-4, mesa 1:26.2.4-1, omacam 0.1.9-1, omamobile-apps 0.2.0-1, omamobile-backlight-oriole 0.1.0-8, omamobile-battery-counters-oriole 0.2.0-1, omamobile-bluetooth-oriole 0.1.2-2, omamobile-bluez 5.87.omamobile1-1, omamobile-charger-cache-oriole 0.1.0-8, omamobile-device-oriole 0.2.0-1, omamobile-firstboot 0.1.1-1, omamobile-gpu-oriole 0.1.0-3, omamobile-keyring 20261013.oriole1-1, omamobile-mirror 0.2.1-1, omamobile-nfc-oriole 0.1.0-9, omamobile-omarchy 0.1.6-1, omamobile-oriole-local-safeguards 0.1.9-2, omamobile-owl-widgets 0.1.1-1, omamobile-shell 0.2.15-1, omamobile-torch-oriole 0.1.1-2, omamobile-wifi-oriole 0.1.8-3, omarchy 4.0.4-2, quickshell 0.3.2.r7.g2b4501a-1

Recover portable helpers and shared policy without enabling hardware services. Keep experimental battery and boot-check services outside this initial package.
@derekross

Copy link
Copy Markdown
Owner

Thank you for starting a public package for the Pixel 6. Keeping it minimal, with nothing enabled at install, is the right instinct. The copied files (reboot-to.c, the storage rule, the NetworkManager drop-in, the licences) match cheetah's byte for byte, tools/pr-check is clean and CI is green. Before it can come out of draft, here's what I'd need:

  1. Run it on the phone. A package change needs a tools/phone-smoke report from the phone it's for (CONTRIBUTING §6.5), and right now the Pixel 6 runs your local omamobile-device-oriole 0.1.9-2 (24 files), not this 0.1.0. Please install this package's head on the phone (tools/phone-install-package packages/omamobile-device-oriole), check pacman -Q afterwards, and paste the report. One version note: pacman counts the installed 0.1.9 as newer than 0.1.0, so the public package would never replace it on an update. Please start above it (0.2.0, say), or say in the README that the local one has to be removed first.

  2. Please leave omamobile-bootctl out for now. It's cheetah's tool, unchanged, and it writes only with --i-understand, but marking slot B on this phone is on hold (device-oriole: record native boot and RTC status finding #200), and how marking should work on the newer phones is open in Marking slot B successful: the plan after #125's hold, and for the Pixel 8 Pro #252. The 9a's package doesn't carry the tool while its marking is on hold either. Please drop the source, LICENSE.Apache-2.0, and its build and test lines, and we'll add them back once Marking slot B successful: the plan after #125's hold, and for the Pixel 8 Pro #252 is settled for the Pixel 6.

  3. Make it a full device package. device/oriole/device.conf names PACKAGE=omamobile-device-oriole, so tools/build-rootfs puts this package into every Pixel 6 image. The image's pacman.conf reads [omamobile] from /etc/pacman.d/omamobile-mirrorlist, and the device package is what ships that file. As it stands, an image built with this package would have no server for our repository. Please carry what every other device package has: the mirrorlist with its pacman install script, depends on omamobile-keyring, and provides/conflicts=(omamobile-device). packages/omamobile-device-panther is a small example to copy from.

  4. The blacklists need their evidence in the repository. The README cites a "recovered Oriole ROOTFS-PLAN coldplug audit", which isn't in the repo. Please add a short NOTES entry (it can go in device-oriole: record native boot and RTC status finding #200) saying what each of the six modules did when it autoloaded, with the kernel release. Also, two of them, max17042_battery and max77759_charger, are the gauge and the charger, while device-oriole: record native boot and RTC status finding #200's matrix row reports a battery level and charging on the same phone. Which drivers produced those readings, and should a public image really block them?

  5. A question about CriticalPowerAction=PowerOff. On the Pixel 7 and 7 Pro, a power-off from Linux restarts the phone (panther NOTES, 2026-10-27). If a Pixel 6 does the same, a flat battery would restart it again and again. Have you seen a real power-off on the Pixel 6 under our kernel? If not, please note it next to the setting.

  6. Smaller: please have test/all compare the files you copied with cheetah's, inside an if [[ -f ../omamobile-device-cheetah/... ]] guard as panther's test does, so any drift shows up in CI.

Nothing here is urgent, and I'm happy to look again whenever you've had a boot with it.

Gr-linkk and others added 5 commits November 3, 2026 15:07
Provide the repository configuration and package metadata, remove the unqualified boot-control tool, and document the safeguards required before replacing a local integration package.
Record the first-light audit without attributing untested behavior to the blocked drivers. Explain the separate experimental battery provider and the limits of charger status evidence.
Record ARM64 build and temporary-directory installation on the phone.
Keep baseline smoke findings distinct from exact-package qualification;
the installed integration safeguards were preserved.

Co-Authored-By: Codex <noreply@openai.com>
Use a version above the installed local device package so normal pacman
checks admit migration with the preserved integration safeguards.

Record native makepkg and signed installation evidence, including the
private companion and outstanding standalone/boot qualification limits.

Co-Authored-By: Codex <noreply@openai.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @Gr-linkk, for working through every point from my last comment. I went through them one by one against the new head:

  • It's installed on the phone, with the smoke report, and at 0.2.0, so it sorts above your local 0.1.9.
  • omamobile-bootctl is gone, and the test makes sure it stays gone.
  • It's a full device package now: the keyring dependency, provides/conflicts, and the mirrorlist and install script, which match panther's byte for byte.
  • The coldplug note says honestly what the audit did and didn't show, and where the battery reading really came from.
  • The power-off caveat is next to the setting.
  • The copied files are compared with cheetah's inside the guard.

test/all passes here on today's main too.

One thing left before it comes out of draft, and it comes from your own notes: on your phone the package runs next to a private companion that keeps the boot check, the battery scripts and the reboot watchdog. device/oriole/device.conf names this package, so an image built for a Pixel 6 would get this package alone, without those. The reboot watchdog can simply come in: cheetah's system.conf.d/10-omamobile-reboot-watchdog.conf (RebootWatchdogSec=1min), compared with cheetah's like the other copies, as long as the watchdog driver binds on your kernel. What does the boot check do? If it marks slot B, it stays out until #200 and #252 settle, and the README should say what happens after three unmarked boots. If it's a hang marker or a start guard, I'd like it here. If you'd rather not bring them in yet, a line in the README and in device.conf saying that an image built from this package isn't a supported install yet would do for now.

One small, optional one: test/all checks the host-root refusal by actually running ./install /. It's safe today because the check comes first, but if that check ever broke, the test would write into the build machine's root. ./install relative-dir hits the same refusal without that risk.

When it merges I'll write the squash title and body myself (omamobile-device-oriole 0.2.0: …), so the merge commit's message doesn't land on main. Thanks for keeping the safety notes so clear.

Separate the private staged-start guard from boot-success marking and
explain why it cannot be copied into this minimal policy package alone.
Mark the descriptor and README as unsupported for standalone installs,
with startup and reboot-watchdog qualification still outstanding.

Exercise install refusal with a relative destination instead of the host
root. No installed policy, services or package version change.

Co-Authored-By: Codex <noreply@openai.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @Gr-linkk. This answers my last point the way I suggested:
the README and device.conf now say up front that an image built from
this package alone isn't a supported Pixel 6 install, and the new notes
explain what the private boot check does, that it doesn't mark slot B,
and what an unmarked slot B does after its tries. Deferring the reboot
watchdog until its binding is checked on a public kernel is fine with me.
Thanks for switching the refusal test away from / too.

I ran test/all on the branch merged with today's main and it passes,
with the copied files still identical to cheetah's. From my side it's
ready whenever you take it out of draft. I'll write the squash title and
body myself at merge.

Two small optional ones, if you're touching it again: the refusal test
would be a little stronger if it also checked the refusal message, so it
can't pass on some other failure; and the paragraph about slot A could
add what to do if stock does come up there: decline any offer to reset,
and take a fresh read-only backup of the radio partitions after any stock
boot.

One heads-up for later: two open PRs (#459 and #403) update every copy of
reboot-to.c. If either lands after this one, I'll make sure it updates
the Pixel 6 copy as well, so your comparison against cheetah's keeps
passing.

@Gr-linkk
Gr-linkk marked this pull request as ready for review November 9, 2026 15:41
derekross added a commit that referenced this pull request Nov 9, 2026
Round 46's confirmed candidates from the reviews of the open
omarchy-mobile and linux-omarchy-mobile pull requests, the round's own
process slips (shared refs and scratch files deleted by glob, a
maintainer fixup that skipped the package's test/all, an amend on a
contributor branch) and the 2026.11.08 release PRs (#547, #548, #549,
#550, #560). Repeats of existing lessons extend them instead: L-45
(kernel #222), L-183 (#409), L-426 (the round's version collisions,
now set by hand at merge) and L-550 (#471 with #420). C-34 and C-35
correct our #290 shell suggestion and round 45's claim that
tests/dipper-probe guards anything.

Stacked on #551 (round 45's lessons).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross added a commit that referenced this pull request Nov 9, 2026
Round 46's confirmed candidates from the reviews of the open
omarchy-mobile and linux-omarchy-mobile pull requests, the round's own
process slips (shared refs and scratch files deleted by glob, a
maintainer fixup that skipped the package's test/all, an amend on a
contributor branch) and the 2026.11.08 release PRs (#547, #548, #549,
#550, #560). Repeats of existing lessons extend them instead: L-45
(kernel #222), L-183 (#409), L-426 (the round's version collisions,
now set by hand at merge) and L-550 (#471 with #420). C-34 and C-35
correct our #290 shell suggestion and round 45's claim that
tests/dipper-probe guards anything.

Stacked on #551 (round 45's lessons).

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Check the staging refusal diagnostic so an unrelated install failure cannot
satisfy the test. Keep the safe relative destination.

Explain that any stock boot requires declining reset offers and fresh
read-only protected radio backups before further radio work.

Validation: package test/all passes; no runtime or version change.
No phone access or hardware qualification.

Co-Authored-By: Codex <noreply@openai.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @Gr-linkk. The last commit takes both optional points: the refusal test now checks the message, and the slot A paragraph says to decline a reset and take a fresh read-only backup after any stock boot. Nothing else changed, the package test passes here on the branch merged with today's main, and the copied files still match cheetah's.

I'm merging it as omamobile-device-oriole 0.2.0, with the scope the README states: a first-light policy package, not yet a supported install on its own. As promised, if #459 or #403 lands later I'll see that the Pixel 6 copy of reboot-to.c moves with them.

@derekross
derekross merged commit fdd3ab3 into derekross:main Nov 11, 2026
2 checks passed
W4HO-ham added a commit to W4HO-ham/omarchy-mobile that referenced this pull request Nov 11, 2026
The Pixel 6's package came to main with its own copy of reboot-to.c (derekross#222), held to the 7 Pro's by its test. This pull request changes that source (the note for a deliberate restart, no printf family), so the ninth copy gets the same file.
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
Round 46's confirmed candidates from the reviews of the open
omarchy-mobile and linux-omarchy-mobile pull requests, the round's own
process slips (shared refs and scratch files deleted by glob, a
maintainer fixup that skipped the package's test/all, an amend on a
contributor branch) and the 2026.11.08 release PRs (derekross#547, derekross#548, derekross#549,
derekross#550, derekross#560). Repeats of existing lessons extend them instead: L-45
(kernel derekross#222), L-183 (derekross#409), L-426 (the round's version collisions,
now set by hand at merge) and L-550 (derekross#471 with derekross#420). C-34 and C-35
correct our derekross#290 shell suggestion and round 45's claim that
tests/dipper-probe guards anything.

Stacked on derekross#551 (round 45's lessons).

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
…or the Pixel 6 (derekross#222)

A device package for the Pixel 6 with nothing enabled at install: the storage udev rule, USB gadget networking, `reboot-to`, the mirrorlist, sleep modes off, and blacklists for six modules whose coldplug isn't qualified. The README and `device.conf` say that an image built from this package alone is not a supported install: the boot check and recovery safeguards of the contributor's private companion are not in it, and slot B is not marked. Installed and smoke-tested on the contributor's Pixel 6 next to that companion.

Co-Authored-By: Codex <noreply@openai.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants