Repository navigation
arm64: dts: exynos: google: Add Pixel 6a (bluejay) first-light board - #179
Conversation
The Pixel 6a (bluejay) is a gs101 (Tensor) board like the Pixel 6 and Pixel 6 Pro, and the binding's description already names it. Add its compatible to the gs101 board list. Assisted-by: Claude Fable 5.1 Signed-off-by: PeterXMR <xmrpeter@proton.me>
|
Thank you, @PeterXMR, and welcome. This is a careful first-light board: the file is oriole's with bluejay's model and compatible, and every value I checked is in the Google sources you link. On my side: our gate passes on today's The one open question is the first boot. Our contributing guide asks for a Two optional things:
Thanks again for the thorough sourcing. (Maintainer: if you'd rather merge it now on the #14 precedent, replace the "first boot" paragraph with "I'm merging this now, as we did for oriole's |
Add a first-light board file for the Pixel 6a: oriole's board file with bluejay's model and compatible, without oriole's two board-specific LDOs. Everything it enables, it enables as oriole's does, through gs101-pixel-common.dtsi: the console UART, USB, both cluster watchdogs, UFS and its PHY, the S2MPG10/11 regulators, the MAX77759 Type-C controller, fuel gauge and PMIC, and the battery EEPROM. The GPU stays disabled, and there is no panel or touch node. bluejay boots the same two gs101 base device trees as oriole and raven: Google's bluejay tree builds every bluejay board overlay on gs101/gs101-a0.dtb and gs101/gs101-b0.dtb, whose soc_id is 0x09845000 and 0x09845001. So this DTB carries the B0 value as oriole's does, and the image tooling writes the A0 copy from it. The panel is a 1080x2400 S6E3FC3 variant (samsung,s6e3fc3_6a in Google's gs101-bluejay-display.dtsi), oriole's size, so the splash carve-out and the simple-framebuffer keep oriole's values. The retail board overlay is MP 1.0, board_id 0x20506 and board_rev 0x10000; the board dtbo for ABL and the image packing live in the omarchy-mobile repository (device/bluejay). Oriole's ldo14m (avdd18_tcxo) and ldo31m (nfc) are left out. Upstream marks both as board specific, and Google's bluejay tree takes both from the shared gs101-pmic.dtsi, regulator-always-on there: L14M_AVDD_TCXO, and L31M, which gs101-whitefin2-pmic.dtsi (included by bluejay and by oriole) renames to L31M_DISP_VDDI. Left undescribed, they stay as the bootloader leaves them, as on raven. Not booted yet: no Pixel 6a has run this tree. Apart from the model, the compatible, the LDO block and its regulator include, the file matches oriole's, its comments on soc_id and the bootargs included; the splash and framebuffer blocks are oriole's text, so its copyright lines stay. The only other differences between the two boards in postmarketOS's gs101-bluejay.dts are ST FTS touch values, for a driver this tree does not have. Link: https://android.googlesource.com/kernel/devices/google/bluejay/+/refs/tags/android-16.0.0_r0.9/dts/gs101-bluejay-mp1_0.dts Link: https://android.googlesource.com/kernel/devices/google/bluejay/+/refs/tags/android-16.0.0_r0.9/dts/gs101-bluejay-display.dtsi Link: https://android.googlesource.com/kernel/devices/google/bluejay/+/refs/tags/android-16.0.0_r0.9/dts/Makefile Link: https://android.googlesource.com/kernel/devices/google/gs101/+/refs/tags/android-16.0.0_r0.9/dts/gs101-b0.dts Link: https://android.googlesource.com/kernel/devices/google/gs101/+/refs/tags/android-16.0.0_r0.9/dts/gs101-pmic.dtsi Link: https://gitlab.postmarketos.org/soc/tensor-exynos/linux/-/blob/e8f42849a581/arch/arm64/boot/dts/exynos/google/gs101-bluejay.dts Assisted-by: Claude Fable 5.1 Signed-off-by: PeterXMR <xmrpeter@proton.me>
e85110d to
93a6438
Compare
|
Thank you for the review @derekross . I've added oriole's soc_id and bootargs comments to gs101-bluejay.dts as suggested; the DTB is byte-identical before and after, and checkpatch reports nothing on the change. On the first boot: I have a Pixel 6a, but I can't run the boot test myself. The s2idle drop-in is noted for the bluejay device package. |
|
Thank you, @PeterXMR. The new head adds oriole's two comments word for word, and nothing else changes in the file, so the DTB is the same as before, as you say. Our gate on today's On the first boot: thanks for saying so plainly. I'm merging this once CI is green, as we did for oriole's |
Add a first-light board file for the Pixel 6a: oriole's board file with
bluejay's model and compatible, without oriole's two board-specific
LDOs. Everything it enables, it enables as oriole's does, through
gs101-pixel-common.dtsi: the console UART, USB, both cluster
watchdogs, UFS and its PHY, the S2MPG10/11 regulators, the MAX77759
Type-C controller, fuel gauge and PMIC, and the battery EEPROM. The GPU
stays disabled, and there is no panel or touch node. The first commit
adds the compatible to the gs101 board binding.
bluejay boots the same two gs101 base device trees as oriole and raven:
Google's bluejay tree builds every bluejay board overlay on
gs101/gs101-a0.dtb and gs101/gs101-b0.dtb, whose soc_id is 0x09845000
and 0x09845001. So this DTB carries the B0 value as oriole's does, and
the image tooling writes the A0 copy from it.
The panel is a 1080x2400 S6E3FC3 variant (samsung,s6e3fc3_6a in
Google's gs101-bluejay-display.dtsi), oriole's size, so the splash
carve-out and the simple-framebuffer keep oriole's values. The retail
board overlay is MP 1.0, board_id 0x20506 and board_rev 0x10000; the
board dtbo for ABL and the image packing live in the omarchy-mobile
repository (device/bluejay).
Oriole's ldo14m (avdd18_tcxo) and ldo31m (nfc) are left out. Upstream
marks both as board specific, and Google's bluejay tree takes both from
the shared gs101-pmic.dtsi, regulator-always-on there: L14M_AVDD_TCXO,
and L31M, which gs101-whitefin2-pmic.dtsi (included by bluejay and by
oriole) renames to L31M_DISP_VDDI. Left undescribed, they stay as the
bootloader leaves them, as on raven.
Not booted yet: no Pixel 6a has run this tree. Every value comes
from Google's GPL bluejay and gs101 device trees (android-16.0.0_r0.9)
or is oriole's; the commit message links each source. Apart from the
model, the compatible, the LDO block and its regulator include, and
oriole's comments, the file matches oriole's, and it keeps oriole's two
copyright lines. The only other differences between the two boards in
postmarketOS's gs101-bluejay.dts are ST FTS touch values, for a driver
this tree does not have. The precedent for merging before a boot is
oriole's soc_id commit (#14), on the same kind of evidence.
Checks, host (dtc 1.8.1, dtschema 2026.9): gs101-bluejay.dtb builds
with -@ from the preprocessed DTS; dt-validate against the tree's
processed schema reports 0 messages, the same as gs101-oriole.dtb;
dt-doc-validate and dt-mk-schema accept google.yaml; checkpatch
--strict: 0 errors, 0 checks, one FILE_PATH_CHANGES warning for the new
file (MAINTAINERS already lists arch/arm64/boot/dts/exynos/google/, as
for the lynx and panther first-light commits); tools/pr-check: no
identifiers, no blobs; the branch is a fast-forward of omamobile. A
pre-submission review against docs/review/LESSONS.md (L-6, L-27, L-29,
L-93, L-100, L-120, L-149, L-187, L-240, L-336) read every Link: at its
tag and found each claim in the message there.
tools/kernel-pr-check --build --merge, in the build container:
The descriptor pull request (device/bluejay: device.conf, notes, the
board overlay, the DEVICES.md row) is derekross/omarchy-mobile#265 and
names this branch; DEVICE=bluejay bin/omamobile images needs this DTB
in a kernel build, so this PR merges first.