Skip to content

arm64: dts: exynos: google: oriole: describe disabled BCM4389 Bluetooth - #159

Merged
derekross merged 1 commit into
derekross:omamobilefrom
Gr-linkk:oriole-bcm4389-disabled
Nov 7, 2026
Merged

derekross merged 1 commit into
derekross:omamobilefrom
Gr-linkk:oriole-bcm4389-disabled

Conversation

@Gr-linkk

Copy link
Copy Markdown
Contributor

Add Oriole's BCM4389 Bluetooth child on APM USI1/UART16, disabled together with its parent controllers pending current-kernel qualification. GPIOs and host-wakeup IRQ follow the retained Oriole bring-up wiring; maximum speed is 3 Mbaud.

Dependencies already present in base 51fb40d2b87398dd38ec860e9cda4e487fbf3b0d:

  • b8c4ca7a6: GS101 APM USI1/UART16 SoC nodes, reused without duplication.
  • b3664451c: corrected read-only APM mux descriptions, not reapplied.
  • Existing BCM4389 binding and hci_bcm support.

The UART baud clock uses CMU_APM CLK_GOUT_APM_APM_USI1_UART_IPCLK, parent dout_apm_usi1_uart, sourced from gout_apm_func. No fixed-factor workaround, assigned rate, or shared mux write is added. No dependency on PR #98 (GS201 Lynx) or #122 (Zumapro Tegu); existing open PRs were checked for overlap.

Validation:

  • Clang arm64 defconfig, targeted Oriole DT build with W=1: pass.
  • Relevant USI, Samsung UART and Broadcom Bluetooth schema validation: pass.
  • git diff --check and strict checkpatch (--no-signoff): pass.
  • Fresh independent read-only review using review-agent: no findings.

Limitations: no phone operations or current-kernel hardware validation. Historical loopback, HCI and BLE keyboard evidence belongs to 4bd35d34abb9-dirty with the historical clock workaround; it does not qualify this clock path or kernel. Firmware/address provisioning, runtime UART/HCI qualification, audio and suspend are outside this disabled integration. The historical enable overlays reference the removed workaround clock and must be adapted before any future trial; no enable overlay is shipped here. Hardware qualification remains the draft's merge blocker.

@Gr-linkk
Gr-linkk force-pushed the oriole-bcm4389-disabled branch from ede647c to 7eb78da Compare October 31, 2026 19:25
@derekross

Copy link
Copy Markdown
Owner

Thanks for this. I've approved the CI runs. I'll review it properly once it's out of draft and the questions on omarchy-mobile#200 are answered.

@Gr-linkk
Gr-linkk force-pushed the oriole-bcm4389-disabled branch from 7eb78da to fc47cf8 Compare November 3, 2026 23:29
@Gr-linkk
Gr-linkk marked this pull request as ready for review November 3, 2026 23:31
@Gr-linkk
Gr-linkk force-pushed the oriole-bcm4389-disabled branch from fc47cf8 to fa104f7 Compare November 3, 2026 23:39
@derekross

Copy link
Copy Markdown
Owner

Thank you, @Gr-linkk, and thanks for keeping it disabled until it's been tried on this kernel. I checked the wiring
against raven's node on omamobile: the same three lines (gpp16-2, gpp14-3, gpa6-6 rising) and 3 Mbaud. Both come
from Google's shared gs101 Bluetooth description, so that part is right.

Two changes before I merge:

  1. Pin states. Raven's node also carries pinctrl-0 = <&bt_reg_on &bt_dev_wake &bt_host_wake>, with the three
    groups in gs101-raven.dts (gpa6-6 as EINT with a pull-down, gpp14-3 and gpp16-2 as outputs with no pull, all
    2.5 mA). Please add the same to oriole. It matters because the node will be enabled later by an overlay, and the
    Samsung pin controller only takes pin groups at its own probe. Groups added by an overlay at runtime are refused,
    so they have to be in the board DTS from the start.
  2. Subject prefix. This file's history uses arm64: dts: exynos: google: gs101-oriole:, and kernel PRs land
    commit by commit, so please use that prefix.

Two optional touches: the status = "disabled" lines on &serial_apm1 and &usi_apm1 repeat what gs101.dtsi
already sets, so the board file only needs the child. And a line in the message saying the wiring matches raven's
node and Google's common gs101 Bluetooth file would help a later reader.

Our gate passes on today's omamobile (clean merge, sign-off, checkpatch, device trees build with no new warnings).
I'll approve the CI run so dtbs_check runs too.

Describe the BCM4389 on the existing APM USI1 UART, including the pin
states required when the device is enabled later. The control lines and
3 Mbaud rate match Raven and Google's common gs101 Bluetooth wiring.

Keep the controller disabled until this kernel and its native CMU_APM
baud clock path have been qualified on Oriole. Historical bring-up used
a different dirty kernel and a clock workaround.

Signed-off-by: Garrett lindquist <gr_link@hotmail.com>
@Gr-linkk
Gr-linkk force-pushed the oriole-bcm4389-disabled branch from fa104f7 to 772d700 Compare November 5, 2026 23:55
@derekross

Copy link
Copy Markdown
Owner

Thank you, @Gr-linkk. Both changes are in, and the optional ones too. The node and its three pin groups now match raven's line for line, the prefix is the file's own, and the board file carries only the child. Our gate passes on today's omamobile: a clean merge, your sign-off, checkpatch with no warnings, and all ten Google device trees building with no new warnings. I'll approve the CI run so dtbs_check runs on this head, and merge once it's green.

One small thing: the PR description still says "hardware qualification remains the draft's merge blocker". That's no longer true, and only the commit message lands, so it's just for anyone reading the PR later. When you get to trying it on the phone, the overlay will need to switch on usi_apm1, serial_apm1 and this child, with a gated runbook in oriole's plans.

@derekross
derekross merged commit fe12fd2 into derekross:omamobile Nov 7, 2026
1 check passed
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