Hi Lucas,

On Thu, Jul 9, 2026 at 8:54 AM Lucas <[email protected]> wrote:
>
> I am trying to use the odroid HC4 with Armbian Linux and two 3.5 inch SATA
> HDDs physically inserted. I believe I have traced a bug to uboot code at
> the start of the boot process.
>
> The issue is documented here:
> https://github.com/armbian/build/issues/9532
>
> I have two versions of Armbian, one works and one doesn't:
> Armbian_26.5.1_Odroidhc4_trixie_current_6.18.33_minimal.img.xz - fails
> Armbian_25.8.1_Odroidhc4_bookworm_current_6.12.41_minimal.img.xz - works
>
> Both versions boot fine with no HDDs inserted. Some SATA devices boot fine.
> These are powerful WD Red enterprise drives. When I interrupt boot by
> pressing any key, I can run `scsi scan` and `boot` and have it boot
> successfully on both versions. I also see HDMI visual artifacts during the
> scsi scan sequence (the text becomes garbled on a few lines) that always
> means the boot will be successful.
>
> I tried compiling edits to /boot/boot.cmd. These edits did persist (I saw
> my echo commands) but did not fix the problem. There is part of the boot
> sequence that happens before the compiled boot.cmd runs.
>
> I tried setting preboot flags without recompilation, but can't figure out
> if this is supported on Armbian / HC4.
>
> I believe this preboot portion is controlled by uboot.
>
> I made some edits to the armbian code and ran a `compile.sh` command that
> seems to create a new uboot. I wrote that to the SD card and the board no
> longer boots with that card.
>
> In order to diagnose further, I could use some help confirming that this
> portion of boot is controlled by uboot.

PCI-based SATA drives could be spinning up either during preboot or
after, depending how u-boot is configured. It would be more helpful if
you can post the following info:

- Which version of u-boot, and whether Armbian has any downstream
patch applied on top of the mainline u-boot source. If you're not
sure, post the GitHub URL of the Armbian u-boot source.
- The content of the defconfig file used in this u-boot, ie.
./configs/xxx_defconfig (where xxx is the board name). It probably can
be found in some build shell scripts.
- If possible, power up, interrupt u-boot countdown and list the following info.
ver
printenv

For reference, here is the commit to fix this problem, but I have
tested only with <= 4TB HDDs.
https://github.com/u-boot/u-boot/commit/53cc4332b3b37218a7cdab8bdb953da57eec2668

All the best,
Tony

Reply via email to