Skip to content

device/akita: stage 0 notes and descriptor for the Pixel 8a - #308

Merged
derekross merged 1 commit into
derekross:mainfrom
aninmucek:device-akita-stage0
Nov 2, 2026
Merged

derekross merged 1 commit into
derekross:mainfrom
aninmucek:device-akita-stage0

Conversation

@aninmucek

Copy link
Copy Markdown
Contributor

What and why

Stage 0 for the Pixel 8a (akita, zuma, Tensor G3), as claimed on #1: the device notes, the descriptor and its empty board overlay, and akita's line in DEVICES.md (CONTRIBUTING §5.1, §6.3 and §6.4; the same shape as #265). Every value comes from a retail 8a with root and the stock CP3A.261005.005 factory image, checked against Google's GPL akita and zuma device trees at android-16.0.0_r0.10. The table at the end of the notes gives each descriptor value's source.

Akita is husky with three board values changed: its bootloader boots the IPOP base DTB (B0/B1 IPOP, pkg_mode 1) where husky's boots FOPLP, its board overlay is dtbo entry 18 (board_id 0x40406, board_rev 0x10000), and its panel is 1080x2400. The descriptor writes no display register (no SIMPLEFB_TRIG_CON, L-4): DECON0's TRIG_CON at handover is still to be read on an 8a. Images wait for the board DTS, zuma-akita.dts, which isn't written yet.

Refs #1

Device and build

  • Phone: Pixel 8a (akita), retail, MP1.0, bootloader unlocked: stock CP3A.261005.005 on both slots, with a Magisk-patched stock init_boot_b for root. No build of ours has been flashed or booted.
  • Kernel release (uname -r): none: there is no akita kernel yet.
  • Packages changed, with versions: none.

Checks (no phone needed)

  • tools/pr-check origin/main passes
  • tests/all passes (260 ok)
  • tests/packages: no package is touched
  • Each changed package has its version bumped: no package changed
  • 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:

Also checked with the loader (L-395): DEVICE=akita bin/omamobile initramfs-config packs the six-name UFS_RW and the USB product "Pixel 8a"; the overlay compiles with dtc -@.

On the phone

  • I ran this on my own phone, booted into this build: nothing here can be built or booted yet. Stage 0 itself ran on my 8a (CONTRIBUTING §5.1): stock facts with root, fastboot getvar all, both slots matched with two full-OTA sideloads, a factory reset, and two checksummed backups of every partition except userdata and super, kept private.
  • What I did not test: no akita DTB or boot images exist, so DEVICE=akita bin/omamobile images can't run until zuma-akita.dts does; VENDOR_CMDLINE is husky's line [I]; the silicon revision (B0 or B1) isn't known.
  • Nothing in this change writes outside slot B and userdata, weakens the storage guard, or reads registers from userspace (AGENTS.md §1): UFS_RW is the five slot-B boot images and userdata, and the descriptor carries no register write or probe.

No tools/phone-smoke report: no build of ours runs on the phone yet.

🤖 Generated with Claude Code

Stage 0 for the Pixel 8a (akita, zuma), from a retail phone with root
and the stock CP3A.261005.005 factory image, checked against Google's
GPL akita and zuma device trees at android-16.0.0_r0.10. Both slots
carry CP3A.261005.005 with bootloader akita-17.0-15681527, and two
checksummed backups of every partition but userdata and super are
kept off the repository.

The descriptor is husky's with akita's board values: the IPOP base DTB
entries (this phone boots B0/B1 IPOP, husky FOPLP), dtbo entry 18
(board id 0x40406, rev 0x10000) and the 1080x2400 panel. It writes no
display register: DECON0's TRIG_CON at handover is still to be read on
an 8a. Images wait for the board DTS, zuma-akita.dts, not written yet.

tests/storage-guard and tests/usb-product now cover the descriptor.

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

Copy link
Copy Markdown
Owner

Hi @aninmucek, and welcome. Thank you for this. It's a very good stage 0: the backups are done before anything else, both slots are on the same build, stock is kept as the way back, and every value in the descriptor has a source in the table at the end of the notes.

What I checked:

  • The descriptor loads. DEVICE=akita bin/omamobile initramfs-config packs the six-name UFS_RW and "Pixel 8a", with nothing refused. tests/storage-guard, tests/usb-product and tests/device-overlays pass on your branch, and the empty overlay compiles with dtc -@. tools/pr-check passes, and the branch merges cleanly onto today's main.
  • The values against Google's GPL trees at android-16.0.0_r0.10. zuma-akita-mp.dts has board_id = <0x40406> and board_rev = <0x010000>, and the PVT board before it is 0x40405, so you used the board ID and not the dtbo index. zuma-b0-ipop.dts has B0_IPOP and B1_IPOP with product_id 0x9865, major 1, minor 0 and 1, and pkg_mode 1, which is exactly your SOC_COMPATIBLE. zuma-akita-display.dtsi has the 1080x2400 mode and the ak3a and ak3b panels, and DECON0 is at 0x19470000 in zuma-drm-dpu.dtsi. The rollback index 1791158400 is 2026-10-05, the build on both of your slots.
  • Safety. UFS_RW is slot B's five boot images and userdata. There's no SIMPLEFB_TRIG_CON, so no display register is written until it's been read on an 8a, and no BRINGUP_PROBE, whose addresses are gs201's. Leaving both unset is right.
  • Nothing identifying in the diff or the description. The SKU is the model number, not a per-phone value.

Nothing here changes any other phone: the shared tests only add akita to their lists.

Two small things, neither blocking:

  • In DEVICES.md, "its bootloader picks the IPOP base DTB where husky's picks FOPLP" reads as true of every 8a. The 8a's factory image carries all four base trees (A0 and B0, FOPLP and IPOP), as husky's does, and we've only seen one 8a, so I'd say "this phone's bootloader picks the IPOP base DTB" until a second 8a confirms it. The notes already say it that way.
  • Husky's descriptor carries short comments next to its values. The table in your notes covers the same ground, so that's fine as it is. A one-line comment at the top pointing at that table would help anyone who opens device.conf first.

Your CI run is waiting for my approval because this is your first PR here, and I'll start it. Once it's green, I'm happy to merge. The next step you describe, a minimal zuma-akita.dts aimed at M1 on linux-omarchy-mobile, sounds right. When you get there, TRIG_CON is read once from our initramfs shell with the bootloader's splash still on the panel, as husky's BRINGUP-M1 test 20 did: devmem 0x19470030, the address only, because a second argument is a write. Please record the value in the notes before SIMPLEFB_TRIG_CON goes in.

Thanks again for such a careful start.

@derekross
derekross merged commit cf019f8 into derekross:main Nov 2, 2026
2 checks passed
@aninmucek

Copy link
Copy Markdown
Contributor Author

Thanks for the careful review and the merge, and for the welcome on #1.

Both small points make sense: "this phone's bootloader" in DEVICES.md until a second 8a confirms it, and a one-line pointer to the provenance table at the top of device.conf. I'll send them as a small follow-up.

Thanks too for the TRIG_CON note. At first light I'll read it once from the initramfs shell with the bootloader's splash still on the panel, devmem 0x19470030 with the address only, and record the value in the notes before SIMPLEFB_TRIG_CON goes in.

@aninmucek aninmucek mentioned this pull request Nov 2, 2026
5 tasks done
derekross pushed a commit that referenced this pull request Nov 3, 2026
derekross's two non-blocking points on #308. DEVICES.md now says this
phone's bootloader picks the IPOP base DTB: the 8a's factory image
carries all four base trees, as husky's does, and only one 8a has been
seen. device.conf gets a one-line pointer to its provenance table in
NOTES.md, for anyone who opens the descriptor first.

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
…s#308)

Stage 0 for the Pixel 8a (akita, zuma), from a retail phone with root
and the stock CP3A.261005.005 factory image, checked against Google's
GPL akita and zuma device trees at android-16.0.0_r0.10. Both slots
carry CP3A.261005.005 with bootloader akita-17.0-15681527, and two
checksummed backups of every partition but userdata and super are
kept off the repository.

The descriptor is husky's with akita's board values: the IPOP base DTB
entries (this phone boots B0/B1 IPOP, husky FOPLP), dtbo entry 18
(board id 0x40406, rev 0x10000) and the 1080x2400 panel. It writes no
display register: DECON0's TRIG_CON at handover is still to be read on
an 8a. Images wait for the board DTS, zuma-akita.dts, not written yet.

tests/storage-guard and tests/usb-product now cover the descriptor.

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
derekross's two non-blocking points on derekross#308. DEVICES.md now says this
phone's bootloader picks the IPOP base DTB: the 8a's factory image
carries all four base trees, as husky's does, and only one 8a has been
seen. device.conf gets a one-line pointer to its provenance table in
NOTES.md, for anyone who opens the descriptor first.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.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