OrangePi 3 LTS: Update U-Boot to v2026.01 and Linux DTS "and other fixups"

[pyavitz]

This PR is meant to fix two issues:

• Ethernet not coming up
• Only one CPU coming up

This also introduces a different way of bringing up bluetooth and drops a dependency on the need for several patches.
Namely Add sunxi-addr driver - Used to fix uwe5622 bluetooth MAC addresses patch and related patches. As these patches are still required by other units, I left them in place.

https://paste.armbian.com/fimarilocu.yaml
Bug report: https://forum.armbian.com/topic/57147-bug-ethernet-rarely-connecting-successfully-in-orange-pi-3-lts/#comment-232222

Summary by CodeRabbit

• New Features

  • Added OrangePi 3 LTS board support with Bluetooth and Wi‑Fi (UWE5622) and systemd service setup
  • Enabled KASLR for enhanced kernel address-space layout randomization
  • Added LED indicators for power and status

• Chores

  • Updated boot configuration, kernel targets, module blacklisting, and device tree hardware descriptions

• Revert

  • Removed an earlier initial device-tree patch entry from the kernel patch series

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds OrangePi 3 LTS support: new board config and U-Boot hooks, wireless setup scripts and systemd service enablement, new/updated device trees for 6.12 and 6.18 (with KASLR/LED changes), kernel/u-boot patches, and removal of an earlier 6.12 DT patch entry.

Changes

Cohort / File(s)	Summary
Board configuration & hooks config/boards/orangepi3-lts.csc	Added BOOT_FDT_FILE, PACKAGE_LIST_BOARD, MODULES_BLACKLIST; updated BOOTCONFIG and KERNEL_TARGET; added post_family_config__use_orangepi3lts_uboot and wireless/systemd hook functions for UWE5622 support.
Wireless helper script packages/bsp/sunxi/sprd-bluetooth	New Bash script that derives a device-based Bluetooth address, writes it to /etc/default/sprd_bt_addr, attaches Sprd HCI UART (hciattach_opi), and unblocks rfkill.
Kernel DT (new 6.12 file) patch/kernel/archive/sunxi-6.12/dt/sun50i-h6-orangepi-3-lts.dts	Added full device tree describing OrangePi 3 LTS hardware (clocks, regulators, LEDs, MMC/SDIO, USB, EMAC, PMIC, Bluetooth, etc.).
Kernel patch removal (6.12 series) patch/kernel/archive/sunxi-6.12/patches.armbian/add-initial-support-for-orangepi3-lts.patch, patch/kernel/archive/sunxi-6.12/series.conf	Removed the previously added 6.12 DT patch and its entry from series.conf (DT deleted and Makefile inclusion reverted).
Kernel DT (6.18 updates) patch/kernel/archive/sunxi-6.18/dt_64/sun50i-h6-orangepi-3-lts.dts	Updated DT: added kaslr-seed, renamed clock node, reworked LED nodes/triggers, adjusted regulators and WiFi power sequencing, changed EMAC timings/phy-mode, reorganized MMC/SDIO nodes, and added Bluetooth under uart1.
U-Boot/kernel config patches patch/u-boot/v2026.01/board_orangepi3-lts/*	Inserted KASLR-related config options (CONFIG_CMD_KASLRSEED, CONFIG_DM_RNG) and added LED-related kernel config entries; adjusted LED labels/default-state in DTS patch.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Poem

🐰 Hops and hops — a board takes shape anew,
DTs and U-Boot patches stitched on cue.
Bluetooth hums, LEDs blink in tune,
I nibble bytes beneath the kernel moon.
🥕

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title references U-Boot v2026.01 and Linux DTS updates, which are primary changes in the PR, though it uses a vague qualifier 'and other fixups' at the end.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 2

🤖 Fix all issues with AI agents

In `@config/boards/orangepi3-lts.csc`:
- Around line 65-70: The hook
post_family_tweaks__enable_uwe5622_wireless_services currently runs
unconditionally and will fail if the unit files aren't present; modify it to
test for each unit file in the chroot before calling chroot_sdcard systemctl
--no-reload enable, e.g. check for the service unit path(s) like
/lib/systemd/system/sprd-bluetooth.service or
/etc/systemd/system/aw859a-wifi.service inside the chroot, and only run
chroot_sdcard systemctl --no-reload enable <unit> when the file exists; if a
unit is missing, call display_alert "$BOARD" "Skipping <unit> enable: unit not
installed" "warning" so enabling is guarded and non-fatal.

In `@packages/bsp/sunxi/sprd-bluetooth`:
- Around line 6-14: The MAC device detection using the sed chain for MACDEV is
brittle (fails for /dev/mapper, overlays); replace it by querying the parent
block device for the root mount via lsblk (e.g., use lsblk -no pkname or -no
NAME for the device mounted at /) to reliably set MACDEV, then attempt to read
/sys/class/block/${MACDEV}/device/cid into HASHDEV and format BTADDR as before;
if the CID file is missing or empty, generate a random locally‑administered
Bluetooth address (set the first octet to 02) and assign it to BTADDR as a
fallback so Bluetooth bring‑up does not abort.

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@patch/kernel/archive/sunxi-6.12/dt/sun50i-h6-orangepi-3-lts.dts`:
- Around line 91-98: The device tree uses CLK_OSC32K_FANOUT in the sdio_pwrseq
node but the header defining that symbol is not included; add the missing
include for dt-bindings/clock/sun6i-rtc.h near the top of the DTS so
CLK_OSC32K_FANOUT resolves when compiling the sdio_pwrseq node.

[github-actions]

✅ This PR has been reviewed and approved — all set for merge!

[github-actions]

✅ This PR has been reviewed and approved — all set for merge!

[jfargen]

The multi-core issue persists even with U-Boot v2026.01:
$ cat /sys/devices/system/cpu/online 0
$ cat /sys/devices/system/cpu/offline 1-3
$ sudo dd if=/dev/mmcblk2 skip=2 count=256 bs=512 2>/dev/null | strings | grep "U-Boot SPL" U-Boot SPL 2026.01_armbian-2026.01-S127a-P2759-H5afb-V7816-B2eb2-R448a (Feb 11 2026 - 04:43:52 +0000)

[pyavitz]

The multi-core issue persists even with U-Boot v2026.01: $ cat /sys/devices/system/cpu/online 0 $ cat /sys/devices/system/cpu/offline 1-3 $ sudo dd if=/dev/mmcblk2 skip=2 count=256 bs=512 2>/dev/null | strings | grep "U-Boot SPL" U-Boot SPL 2026.01_armbian-2026.01-S127a-P2759-H5afb-V7816-B2eb2-R448a (Feb 11 2026 - 04:43:52 +0000)

Thanx for the feedback.

Is this an IMG you built ur self?

I ask because I've run several tests. When using https://github.com/armbian/os to build an IMG I get the multi-core issue, when I force a clean build using armbian/build locally the core issue isn't present. I'm at a bit of a loss and extra eyes on the issue would be very helpful.

[pyavitz]

@jfargen

Try this img:
https://github.com/pyavitz/binary/releases/download/images/Armbian-unofficial_26.02.0-trunk_Orangepi3-lts_trixie_current_6.12.71.img.xz

I removed crust firmware support. It booted as it should in my case.

[jfargen]

@pyavitz I've tested the image and the 4 processors are now showing online.
# cat /sys/devices/system/cpu/online 0-3

[pyavitz]

@pyavitz I've tested the image and the 4 processors are now showing online. # cat /sys/devices/system/cpu/online 0-3

Thanx for getting back.

So this is the dilemma. I run tests, I make slight changes to the boot process; extlinux instead of boot.scr, crust vs no crust, defconfig adjustments, etc... and all of sudden it boots properly. Hence the reason for some of the changes I've made recently, but I feel like something is still missing. Almost seems like some kind of artifact is infecting the build? Because once I do the PR and then try again the CPU bug comes back.

What I'm saying is this; I can remove crust from the build in Armbian, but I'm skeptical of that actually fixing anything.