orangepi5-plus: u-boot: use mainline ATF; bump to v2026.01; drop funky CE config options

[rpardini]

• 🌱 also drop BOOT_SUPPORT_SPI & BOOT_SPI_RKSPI_LOADER as those are for vendor u-boot, not used here

Summary by CodeRabbit

• Updates

  • Updated U-Boot version to 2026.01
  • Refined boot configuration for Orange Pi 5 Plus

• Changes

  • Switched boot scenario to the new mainline/ATF approach
  • Removed SPI boot support options
  • Adjusted U-Boot binary mapping and write behavior for the board

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

[coderabbitai]

📝 Walkthrough

Walkthrough

Board configuration for Orange Pi 5 Plus switched boot scenario to "tpl-blob-atf-mainline", removed SPI-specific variables, bumped mainline U-Boot/ATF from v2025.10 to v2026.01, replaced BL31 blob path with fixed bl31.elf, removed an SPI-specific helper, and inlined internal write_uboot_platform functions.

Changes

Cohort / File(s)	Summary
Orange Pi 5 Plus boot config config/boards/orangepi5-plus.conf	BOOT_SCENARIO changed to "tpl-blob-atf-mainline"; removed BOOT_SUPPORT_SPI and BOOT_SPI_RKSPI_LOADER; BOOTBRANCH and BOOTPATCHDIR updated from v2025.10 to v2026.01; UBOOT_TARGET_MAP now uses BL31=bl31.elf; removed post_config_uboot_target__extra_configs_for_rock5b_mainline_environment_in_spi(); added internal write_uboot_platform and write_uboot_platform_mtd definitions.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• rock-5b: u-boot: edge: bump to v2026.01-rc2; use mainline ATF; bootorder; fancy u-boot #8949 — Mirrors mainline U-Boot/ATF bump and BL31 mapping to bl31.elf.
• rock-4se: binman boot scenario with mainline ATF #8449 — Similar change switching Rockchip boards to mainline-ATF binman flow and BL31 handling.
• radxa-cubie-a5e: maint and fixes #9152 — Related adjustments to U-Boot/ATF branch variables and boot binary write logic.

Suggested reviewers

• igorpecovnik
• amazingfate
• paolosabatino
• pyavitz

Poem

🐰 I nibble code beneath the moonlit tree,
bl31 tucked snug for booting free,
SPI paths vanish, tags move on,
mainline steps where blobs belong.
Hoppity hops—v2026, whee! 🥕

🚥 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 clearly summarizes the three main changes: switching to mainline ATF, bumping u-boot version to v2026.01, and removing SPI configuration options.
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.}

---

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.}

[EvilOlaf]

I don't have a spare 5+ to test this.
Also would be good if, as mentioned, hdmirx on vendor is tested before and after this change.

[rpardini]

hdmirx on vendor

won't work with this, most likely. rkbin blob is required to sneakily remap an interrupt for vendor hdmirx (which is incompatible with mainline hdmirx). not specific to opi either.

[Marsupilami23]

Can confirm, patch works for Orange Pi 5 Plus using BRANCH="current" (6.18.6) and BOARD="orangepi5-plus".

[github-actions]

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

[EvilOlaf]

Does hdmirx in mainline work already? I mean as mentioned this would sacrifice this feature when being on vendor kernel which, sadly, has still some relevance in certain scenarios.

[rpardini]

Does hdmirx in mainline work already?

Yes, since 6.18: torvalds/linux@ccc7f16 - but only with mainline ATF.

[rpardini]

Either way, the point here is not hdmirx, but actually fixing the non-bootness problem reported by @Marsupilami23 (and confirmed fixed).

[EvilOlaf]

right, then since confirmed working, lets do it

[EvilOlaf]

There are a quite a bunch of other boards still using spl blobs. Should this be ajdusted to other boards too? Like rock5b, opi5 (without b or pro or plus or ultra or monsterxxl or whatever ridicoulos name comes up next)?

[rpardini]

There are a quite a bunch of other boards still using spl blobs. Should this be ajdusted to other boards too? Like rock5b, opi5 (without b or pro or plus or ultra or monsterxxl or whatever ridicoulos name comes up next)?

I suggest this:

• first priority is moving boards that use vendor u-boot to mainline (to get rid of the Jammy-only problem)
• then ofc mainline kernel, with correctcly-described hdmirx
• then boards that already use mainline u-boot for everything can be moved to mainline atf

this one I think checks all three.

[igorpecovnik]

build breaks on jammy and noble

[rpardini]

build breaks on jammy and noble

Can't find jobs failing? care to share?

[igorpecovnik]

https://github.com/armbian/os/actions/runs/21265692496/job/61205323777

https://paste.armbian.com/ducotixefu

[rpardini]

armbian/os/actions/runs/21265692496/job/61205323777

paste.armbian.com/ducotixefu

Oh I see, this is vendor u-boot. I was under the impression that this board only had mainline u-boot, even for vendor branch. Let me check.

[rpardini]

@igorpecovnik fixed at #9287

There are a quite a bunch of other boards still using spl blobs. Should this be ajdusted to other boards too? Like rock5b, opi5 (without b or pro or plus or ultra or monsterxxl or whatever ridicoulos name comes up next)?

I suggest this:

• first priority is moving boards that use vendor u-boot to mainline (to get rid of the Jammy-only problem)
• then ofc mainline kernel, with correctcly-described hdmirx
• then boards that already use mainline u-boot for everything can be moved to mainline atf

this one I think checks all three.

I was wrong about this. This board still has vendor u-boot for the vendor branch. Vendor hdmirx should work as long as vendor u-boot actually boot the machine. For users: keep in mind mixing blobs is problematic (SPI vs eMMC vs SD). When trying to boot current/edge (specially from SD card), ensure you don't have a leftover vendor stuff in eMMC/SPI.