Skip to content

Pixel 8a (akita): the first-light board file, with USB, SSH and the boot framebuffer - #192

Merged
derekross merged 3 commits into
derekross:omamobilefrom
aninmucek:omamobile-zuma-akita-dts
Nov 5, 2026
Merged

derekross merged 3 commits into
derekross:omamobilefrom
aninmucek:omamobile-zuma-akita-dts

Conversation

@aninmucek

@aninmucek aninmucek commented Nov 3, 2026 •

Copy link
Copy Markdown
Contributor

Part of derekross/omarchy-mobile#1 (the zuma phones). This adds a board file for the Pixel 8a (akita), on zuma.dtsi, in three commits:

  • dt-bindings: arm: google: Add Pixel 8a (akita): google,zuma-akita in the zuma enum.

  • arm64: dts: exynos: google: Add Pixel 8a (akita) first-light board: husky's first-light file with:

    • akita's model and compatible;
    • the /soc_compatible children its bootloader matches, B0_IPOP and B1_IPOP (product id 0x9865, pkg_mode 1; husky's are FOPLP);
    • husky's bootargs;
    • the DWC3 and its two PHYs switched on.

    Symbols are kept (-@) for the bootloader's board overlay.

  • arm64: dts: exynos: google: zuma-akita: Add the boot framebuffer: simple-framebuffer at 0xfac00000, 1080x2400 a8r8g8b8, with a no-map carve-out, as husky's board file has.

Where the values come from:

  • the 8a's live device tree and fastboot getvar all (a retail phone on stock CP3A.261005.005, bootloader akita-17.0-15681527);
  • Google's GPL trees: akita's zuma-akita-mp.dts, zuma's zuma-b0-ipop.dts, and the ak3b panel driver, which sets MIPI_DSI_MODE_VIDEO in no mode, so the panel runs in command mode;
  • three registers read once each on the booted phone, all three husky's values:
    • DPP0 RDMA_BASEADDR_P0 = 0xfac00000;
    • DECON0 GLOBAL_CON = 0x133;
    • TRIG_CON = 0x3070, the trigger masked.

The full record is in omarchy-mobile's device/akita/NOTES.md and in BRINGUP-M1.md, which comes in derekross/omarchy-mobile#414.

Tested on a Pixel 8a. The kernel was 7.3.0-rc6-omamobile-push1-gce160abd9ffc-dirty: these three commits' content on 4cf2c8b, with the framebuffer commit not yet committed. Since then the base has gained 8 commits, and none of them touch a zuma file. The three files here are byte-identical to what ran.

  • Boot:
    • The bootloader boots the tree from slot B. Only the five *_b boot images were written, by bin/omamobile flash.
    • All nine CPUs come up, with 7.3 GiB of RAM.
    • The watchdog driver stops both cluster watchdogs.
    • The kernel reports no reserved-memory overlap.
    • The booted tree adds only the bootloader's usual nodes: the DRAM banks, its four reservations, kaslr-seed and the initrd range.
  • USB and reboot: the initramfs brings up the USB gadget and SSH 0.4 s into boot. reboot-to bootloader reaches fastboot in 25 s, thanks to reboot=warm.
  • The framebuffer: simpledrm binds, and fbcon runs at 67x75. Once SIMPLEFB_TRIG_CON unmasks DECON0's trigger, the panel shows Omarchy's splash in place and in the right colours, and then console text written over it.
  • The boot log's only warnings are husky's: the eUSB2 PHY's two missing supplies (dummy regulators, also dtbs_check's two warnings) and exynos-clkout … -22.

Not in this PR: UFS, the buttons, the PMICs, touch and the MAX77759, each of which follows once it has run on the phone. Whether this 8a's silicon is B0 or B1 is still open, since zuma has no chip-id driver here yet.

Checks: tools/kernel-pr-check --build OUT --merge origin/omamobile passes everything:

  • a clean merge onto fe9627c;
  • the sign-offs: 3 commits, each signed off by its author, none by an AI;
  • checkpatch --strict with no errors. Its two warnings are both on the first-light commit:
    • MAINTAINERS already covers arch/arm64/boot/dts/exynos/google/;
    • the bootargs line is 204 columns, the same line as in husky's board file;
  • the 10 Google device trees with W=1 and dtbs_check, with no new warnings. The 8a's two eUSB2-supply messages are the ones husky has at the base;
  • dt_binding_check on google.yaml.

🤖 Generated with Claude Code

The Pixel 8a (akita) is a zuma (Tensor G3) board like the Pixel 8 Pro
(husky). Add its compatible to the zuma board list.

Assisted-by: Claude Opus 5.5
Signed-off-by: DarekM <149845949+aninmucek@users.noreply.github.com>
Add a first-light board file for the Pixel 8a: husky's first-light file
with akita's model, compatible and /soc_compatible children, and USB.
zuma.dtsi already describes the first DRAM bank (ABL adds the others),
the fixed reservations, the watchdogs, and the PMU with its reboot
mode; this file switches on the DWC3 and its two PHYs and carries
husky's bootargs.

akita's ABL picks the base DTB by /soc_compatible, as husky's does, but
a retail 8a has the IPOP package: its live tree's children are B0_IPOP
and B1_IPOP (product id 0x9865, pkg_mode 1), the two that Google's
zuma-b0-ipop.dts carries, where husky's are FOPLP.

Every fixed DRAM reservation in the 8a's live tree lies inside one of
zuma.dtsi's, at husky's addresses, except the four that ABL inserts
itself (pkvm_guest_firmware, sec_dram, sec_pt and
debug_kinfo_reserved). The DWC3, the USB DRD PHY, the UART, the PMU and
the watchdogs are at zuma.dtsi's addresses.

There is no framebuffer yet: the splash address is read from DPP0 on a
running 8a before a carve-out goes in, as on husky. UFS, the buttons,
the PMICs, touch and the MAX77759 follow, each once it has run on the
phone.

Tested on a Pixel 8a (bootloader akita-17.0-15681527): the bootloader
boots this tree from slot B, all nine CPUs come up, the watchdog
driver stops both cluster watchdogs, and the initramfs brings up the
USB gadget and SSH 0.4 s into boot. The booted tree has the
bootloader's DRAM banks and its four reservations added, and the
kernel reports no reserved-memory overlap.

Link: https://android.googlesource.com/kernel/devices/google/akita/+/refs/tags/android-16.0.0_r0.10/dts/zuma-akita-mp.dts
Link: https://android.googlesource.com/kernel/devices/google/zuma/+/refs/tags/android-16.0.0_r0.10/dts/zuma-b0-ipop.dts
Assisted-by: Claude Opus 5.5
Signed-off-by: DarekM <149845949+aninmucek@users.noreply.github.com>
ABL hands over with its splash still scanned out: on a booted Pixel 8a,
DPP0's RDMA_BASEADDR_P0 (0x19900040) reads 0xfac00000, as on the Pixel
8 Pro. Describe it as a simple-framebuffer at the panel's native
1080x2400 (a8r8g8b8), with a no-map carve-out so the kernel leaves it
alone, as the Pixel 8 Pro's board file does and the 1080x2400 Pixel 7
and 7a ones do.

The ak3b panel is in DSI command mode (no mode in Google's panel driver
sets MIPI_DSI_MODE_VIDEO), and ABL leaves DECON0's hardware trigger
masked (TRIG_CON 0x19470030 reads 0x3070, as on the Pixel 8 Pro), so
frames reach the panel only once userspace unmasks it.

Tested on a Pixel 8a: simpledrm binds to the framebuffer and fbcon runs
at 67x75. With the trigger unmasked, the panel shows what is drawn
there, in place and in the right colours, and keeps following later
writes.

Link: https://android.googlesource.com/kernel/devices/google/akita/+/refs/tags/android-16.0.0_r0.10/display/panel-google-ak3b.c
Assisted-by: Claude Opus 5.5
Signed-off-by: DarekM <149845949+aninmucek@users.noreply.github.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @aninmucek. This is a tidy board file: husky's first-light file with only what the 8a changes, each value traced to the live tree, Google's GPL trees or a register read on the phone. The framebuffer is in its own commit, after the address was read, as husky's was.

What I checked:

  • Our gate passes on today's omamobile (fe9627c): a clean merge, the sign-offs on all three commits, checkpatch --strict with no errors (the two warnings are the MAINTAINERS note and the bootargs line, the same as husky's), and the 10 Google device trees with W=1 and no new warnings.
  • I ran dt-validate on the merged zuma-akita.dtb. The only messages are the two eUSB2-PHY supplies, the same as husky's at the base. Against the base's schema the same DTB is rejected on its compatible, so your binding commit is what makes it pass. google.yaml itself validates.
  • The Pixel 7 Pro's and husky's DTBs are byte-identical before and after.
  • The carve-out is three cells (#size-cells = <1>), and USB being enabled in the first commit fits CONTRIBUTING's exception for nodes already proven on a same-SoC phone.

CI hasn't run yet because it's waiting for my approval; I'll approve it. The only other thing is the source check for the framebuffer commit that I asked for on omarchy-mobile#414, since that boot was -dirty. Once that's in and CI is green, I'll merge this, then #414.

aninmucek added a commit to aninmucek/omarchy-mobile that referenced this pull request Nov 3, 2026
From the review of derekross#414:

- The status line names derekross/linux-omarchy-mobile#192 and its
  commits. M1.3's kernel commit, ce160abd9, is derekross#192's 1253914a9 with
  the same files. M1.4's -dirty run is derekross#192's 6a2999804: the kept DTS
  is blob 381ad1e0 and the flashed DTB has sha256 9a5d92a5, as derekross#192's.
- Risks: every unmarked boot uses one of slot B's tries. If stock boots,
  decline any factory reset and take a fresh backup.
- M1.2 lists the fastboot getvar values instead of a local helper, and
  the two rollbacks pin the serial.
- NOTES.md: the COMPATIBLE and DTB rows point at derekross#192.

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

Copy link
Copy Markdown
Contributor Author

Thanks for checking it so closely, and for approving CI. The source check is in derekross/omarchy-mobile#414 now, and it matches. The DTS kept with the -dirty M1.4 run is blob 381ad1e0e7cbf00a74a8d59f1cf48af97046adcb, and its DTB (also the one inside the flashed vendor_kernel_boot) has sha256 9a5d92a580a8275d39a67b3a13571532374c85d37c8030111d026257d91a123d, as in your build. The M1.3 boot's commit, ce160abd9, has the same files as 1253914a9 here.

@derekross

Copy link
Copy Markdown
Owner

Thanks, @aninmucek. The source check matches. Your base has moved by one lynx commit since my last look, so I ran the gate again on today's omamobile. It merges cleanly, everything passes as before, and the merged zuma-akita.dtb is still 9a5d92a5…, the one you flashed. The Pixel 7 Pro and husky DTBs are unchanged. I'm sorry CI is still waiting on me; I'll approve it now and merge once it's green, then omarchy-mobile#414.

@derekross
derekross merged commit 0e2739a into derekross:omamobile Nov 5, 2026
1 check passed
@aninmucek
aninmucek deleted the omamobile-zuma-akita-dts branch November 6, 2026 05:13
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