Skip to content

device-panther 0.1.6: AVB footers on boot and init_boot - #396

Merged
derekross merged 4 commits into
derekross:mainfrom
nesmeck:panther-avb-footers
Nov 4, 2026
Merged

derekross merged 4 commits into
derekross:mainfrom
nesmeck:panther-avb-footers

Conversation

@nesmeck

@nesmeck nesmeck commented Nov 3, 2026

Copy link
Copy Markdown
Contributor

What and why

Fixes #342. Slot B images for the Pixel 7 now carry AVB hash footers on boot and init_boot, as the Pixel 7 Pro's do (issue #11).

  • The cause. Stock's vbmeta on the Pixel 7 chains boot (rollback index location 2) and init_boot (location 4); I checked the factory image's vbmeta.img with avbtool info_image. libavb reads a chained partition with no footer as invalid metadata, which an unlocked bootloader doesn't let through. So on a phone whose vbmeta_b is stock's, slot B never starts. angelrenefm's unit, with a full OTA on both slots, showed it: a black screen, no USB, and stock recovery on slot A after about 72 s.
  • Why my unit booted without them. It ran LineageOS before the port. My stock install wrote only the bootloader and radio to slot B, so vbmeta_b doesn't chain them.
  • The change.
    • device/panther/device.conf gets cheetah's three keys, with the values from the factory image's own boot.img and init_boot.img footers. Both units run that build (CP3A.261005.002.A1) on slot A, and the values match OS_PATCH_LEVEL:
      AVB_FOOTERS="boot:0x4000000 init_boot:0x800000"
      AVB_ROLLBACK_INDEX=1791158400
      AVB_SECURITY_PATCH=2026-10-05
      
      The sizes are getvar partition-size:boot_b and init_boot_b.
    • The package's copy of the descriptor follows, so omamobile-boot-update writes the footers on the phone too.
    • device/panther/NOTES.md records the finding and both units' results.

The version. This is device-panther 0.1.6, and omarchy-mobile#281 is also set to ship as 0.1.6 when it merges. Whichever goes in second should be 0.1.7. Edits are allowed if you'd rather do that here.

Device and build

  • Phone: Pixel 7 (panther), two units
  • Kernel release (uname -r):
    • mine: 7.3.0-rc6-omamobile-push1-gdb341d1e29e7;
    • angelrenefm's: 7.3.0-rc6-omamobile-push1-g4cf2c8be1705.
  • Packages changed, with versions: omamobile-device-panther 0.1.6 (descriptor only)

Checks (no phone needed)

  • tools/pr-check upstream/main passes
  • tests/all passes. tests/images-layout covers AVB_FOOTERS, and the package's test checks that its descriptor matches device/panther/device.conf.
  • tests/packages passes
  • Each changed package has its version bumped, and the new version is in the commit subject
  • Commits have area prefixes and bodies that say why; any AI help is a Co-Authored-By: or Assisted-by: trailer, never a Signed-off-by:

On the phone

  • I ran this on my own phone, booted into this build
  • What passed:
    • Bytes. Rebuilt from the same kernel output, vendor_boot, vendor_kernel_boot, dtbo and dtb.img are byte for byte the same, and boot and init_boot match up to their padding and footer. Their footers carry rollback index 1791158400, security_patch 2026-10-05 and os_version 17, padded to 64 and 8 MiB, as stock's do.
    • My unit (vbmeta_b not chaining them): bin/omamobile flash of those images. It booted as before: the same kernel, systemctl is-system-running running, and one retry spent. tools/phone-smoke is below.
    • angelrenefm's unit (stock vbmeta_b, #342): his tree with 81b8d63 on top. bin/omamobile images printed both footer lines, then bin/omamobile flash. The USB gadget appeared 24 s after fastboot dropped off and stayed, bound as cdc_ncm. Before this, nothing reached USB in 72 s.
  • What I did not test:
    • an install from scratch with tools/install-phone with this change (angelrenefm's userdata was already installed);
    • a kernel update written on the phone by omamobile-boot-update;
    • SSH and phone-smoke on angelrenefm's unit, which he'll post on Pixel 7 (panther): a second unit available for testing #342.
  • Nothing in this change writes outside slot B and userdata, weakens the storage guard, or reads registers from userspace (AGENTS.md §1)
tools/phone-smoke report

The two not ok lines don't come from this change:

  • The dmesg lines are from my local test kernel (a BLE mouse, and clkout).
  • omamobile-bt-addr@hci0:N started for the Bluetooth connection devices of yesterday's headset test. The udev rule's hci[0-9]* also matches hci0:N. The script refuses them ("not a controller"), so nothing changes, but the units show as failed. The rule is the same in the cheetah, raven and tegu packages; I'll send a separate fix.

phone-smoke: panther

Kernel 7.3.0-rc6-omamobile-push1-gdb341d1e29e7, tools at df53d0d, 2026-11-03T13:42Z.

Check Result Detail
boot (required) ok kernel 7.3.0-rc6-omamobile-push1-gdb341d1e29e7, slot b, up 34376s
storage guard (required) ok 55 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, 19 error lines in dmesg (15 known, 4 new)
new dmesg errors not ok hid-generic N:N:N: device has no listeners, quitting x1; exynos-clkout exynos-clkout: probe with driver exynos-clkout failed with error N x1; exynos-clkout exynos-clkout: cannot match parent device x1
systemd units not ok failed: omamobile-bt-addr@hci0:11.service omamobile-bt-addr@hci0:6.service
USB gadget (required) ok bound to 11210000.usb
display ok card2-DSI-1 connected, 1080x2400
GPU ok render node on panthor
session (required) ok Hyprland and the Omarchy shell running
lock ok PAM file present, the shell can lock (passwordPam true, locked true)
touch ok FocalTech FT3658
buttons ok gpio-keys
Wi-Fi skip no wireless interface
Bluetooth ok hci0, powered true
battery ok 98% Discharging, 24 °C
sound cards skip none
sensors skip no sensor proxy
cameras skip no libcamera tools
thermal ok hottest zone 41 °C

Packages: hyprland 0.56.2-5.1, libcamera 0.7.2-4.21, mesa 1:26.2.4-1.4, omacam 0.3.3-1, omamobile-apps 0.2.0-1, omamobile-device-panther 0.1.7-1, omamobile-firstboot 0.1.2-1, omamobile-keyring 20261026-1, omamobile-mirror 0.1.1-1, omamobile-omarchy 0.1.12-1, omamobile-owl-widgets 0.1.1-1, omamobile-shell 0.2.28-1, omarchy 4.0.6-1, quickshell 0.3.2.r7.g2b4501a-1

🤖 Generated with Claude Code

nesmeck and others added 3 commits November 2, 2026 18:29
Stock's vbmeta on the Pixel 7 chains boot and init_boot, as on the Pixel
7 Pro (the factory image's vbmeta.img). On a phone whose vbmeta_b is
stock's, footer-less slot B images never start: libavb reads a chained
partition with no footer as invalid metadata, which an unlocked
bootloader doesn't let through. A second Pixel 7, with a full OTA on both
slots, showed it (issue derekross#342): a black screen, no USB, and stock recovery
on slot A after about 72 s. The first unit boots without footers because
its vbmeta_b dates from the OS it ran before.

So the descriptor gets cheetah's AVB_FOOTERS, padded to getvar's
partition sizes, with stock CP3A.261005.002.A1's rollback index and
security patch (its own boot and init_boot footers, matching
OS_PATCH_LEVEL). The package's copy of the descriptor follows, so
omamobile-boot-update writes the footers too.

Rebuilt from the same kernel output, vendor_boot, vendor_kernel_boot,
dtbo and dtb.img are byte for byte the same, and boot and init_boot match
up to their padding and footer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @nesmeck, and thanks to @angelrenefm for the test. The diagnosis is solid: the factory vbmeta.img chains boot and init_boot, it's exactly the 7 Pro's issue #11, and the evidence is a clean A/B. On the second unit, the same kernel and the same initramfs with only the footers added went from nothing on USB in 72 s to a gadget at 24 s. The comment in the descriptor explains the case where the footers aren't read, too.

Checks: tools/pr-check passes, CI is green, and the branch merges cleanly onto today's main. The two device.conf copies are identical. The index (1791158400, 2026-10-05) and the patch level are slot A's build on both your phones, and your comment carries the rule we keep for cheetah: a pull request never raises them. A future owner on an older stock build gets a lower value, not this one.

On the version: I'll merge this one first as 0.1.6, and #281 then becomes 0.1.7. Its version file reads 0.1.5 at the moment, which main already has.

Two small questions, neither blocking:

Thanks for turning this around within a day of the report.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@nesmeck

nesmeck commented Nov 3, 2026

Copy link
Copy Markdown
Contributor Author

Thanks! On the installed package: omamobile-device-panther 0.1.7-1 is a local build from my own branch, ahead of both PRs. It has the charger's modprobe options and the Bluetooth address unit, which are still waiting their turn, and it's neither #281 nor this branch. The footer images came from bin/omamobile images and were flashed directly, so the package version on the root didn't take part in the test.

I've added angelrenefm's smoke report to the second unit's bullet in NOTES (7010b41).

🤖 Generated with Claude Code

@derekross
derekross merged commit e604735 into derekross:main Nov 4, 2026
2 checks passed
derekross added a commit to angelrenefm/omarchy-mobile that referenced this pull request Nov 5, 2026
derekross added a commit to angelrenefm/omarchy-mobile that referenced this pull request Nov 5, 2026
@derekross

Copy link
Copy Markdown
Owner

Thank you, @nesmeck, for the answer and the NOTES line. That settles both questions: the 0.1.7 was your local build and didn't take part in the test, and the second unit's smoke report is now in the record. The new commit touches only that line, CI is green, and the branch still merges cleanly onto main.

Merged as 0.1.6, ahead of the other Pixel 7 pull requests (#281 becomes 0.1.7, as you've already set it). Thanks again.

robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
Stock's vbmeta on the Pixel 7 chains boot and init_boot, as on the Pixel
7 Pro. On a phone whose vbmeta_b is stock's, footer-less slot B images
never start: a second Pixel 7 with a full OTA on both slots showed a
black screen, no USB and stock recovery on slot A after about 72 s
(issue derekross#342). The descriptor gets cheetah's AVB_FOOTERS, padded to
getvar's partition sizes, with stock CP3A.261005.002.A1's rollback index
and security patch, and the package's copy follows, so
omamobile-boot-update writes the footers too. Rebuilt from the same
kernel output, the other images are byte for byte the same. The footer
images boot on both units (NOTES, with the second unit's smoke report).
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.

Pixel 7 (panther): a second unit available for testing

2 participants