Re: Can't Boot Guix on RockPro64

[Ashish SHUKLA]

El 2026-09-09 00:54, mail--- via escribió:

Hi Vagrant!

Thanks for the help!

Unfortunately, I don't think the extra kmods seem to have changed the situation.

I'm attaching the new config.scm and the new boot log.

I don't believe the lines under "loading kernel modules..." were there last 
time, so I think the new kmods are being successfully loaded.

But you can see at the bottom that the emmc still isn't showing under /dev. I 
don't have a lot of experience with bootloader shells like like this, but I'm 
assuming I'd see something conspicuous there if the partitions were visible.

I agree, I should reference something more concrete like a UUID. But there was 
some confusion in the IRC which of the two UUIDs shown with blkid that I should 
be using. And I couldn't get a boot using either of them. So I've been sticking 
with the /dev path.
Very much appreciate the help!

- Dakota

If Debian boots fine on that computer, and the same device gets detected there, then you 
can identify the responsible driver module through sysfs, and also preload it in 
`initrd-modules'. e.g. for my nvme0 device, "nvme" module is responsible.

  ❯ readlink -f /sys/class/nvme/nvme0/device/driver/module
  /sys/module/nvme

You have to locate the apropriate sysfs path though,

If this doesn't work, you can load all the currently loaded modules (in Debian) 
to `initrd-modules'.

Sep 8, 2026, 4:37 PM by vagr...@debian.org:

On 2026-09-08, mail--- via wrote:

I have a Pine64 RockPro64 single-board ARM computer. I am trying to
get Guix running on it, and running into an odd issue. The folks in
the IRC chat suggested I reach out to the mailing list for help.

I am able to run Debian on an SD card and install Guix there. I have
an EMMC module (linked below) that I can attach to the board while
Debian is running.

https://pine64.com/product/128gb-emmc-module/

I am able to successfully partition the module, mount the ext4
partition and run 'guix system init' to install.

Starting the board with only the EMMC module attached, the bootloader
starts successfully, but fails to find the installation
partition. Running 'ls /dev' in the bourneish shell doesn't show any
mmc devices.

I tried looking through dmesg in Debian and adding any kernel modules
that seemed relevant to initrd-modules, but got the same result.

Probably just did not add enough modules ... there are some obscure ones
such as clocks and regulators that are indirect dependencies to get the
hardware fully functional...

Here are the ones I use, which might be more than you need and although
I am booting from SCSI... pretty sure I have all the modules that used
to successfully work on microSD, which should be almost the same as
eMMC... so many tricky combinations!

 (initrd-modules
 (append
 (list
 ;; scsi modules
 "ahci"
 "libata"
 "sd_mod"
 "scsi_common"
 ;; "t10_pi" ; not present in 6.13.x?
 ;; regulators and clocks
 "rk808-regulator"
 "clk-rk808"
 "fixed"
 "fan53555"
 ;; "rk808" ; not present in 6.13.x? replaced by MFD_RK8XX_I2C a.k.a. rk8xx-i2c
 "rk8xx-i2c" ; replaces rk808
 "i2c-rk3x"
 "pl330"
 "dwc3"
 "rtc-rk808"
 "sdhci"
 "sdhci-pltfm"
 "dw_mmc"
 "dw_mmc-pltfm"
 ;; "dw_mmc-rockchip"
 "phy_rockchip_pcie"
 "pcie_rockchip_host"
 "nvme")))

I've also tried to modify the SCM to look for the partition by UUID instead of 
path, with no success.
I'm attaching my config.scm and the log from u-boot.
Any help would be appreciated!

Yeah, you definitely want to use some sort of unique identifier for the
device, e.g. uuid or partition label...

Good luck!

Starting kernel ...

GC Warning: pthread_getattr_np or pthread_attr_getstack failed for main thread
GC Warning: Could not open /proc/stat
Welcome, this is GNU's early boot Guile.
Use 'gnu.repl' for an initrd REPL.

loading kernel modules...
e2fsck: No such file or directory while trying to open /dev/mmcblk1p1
Possibly non-existent device?
File system check on /dev/mmcblk1p1 failed

Curiously, /dev/mmcblk1 is microSD on mine ... do you have a
/dev/mmcblk0 ?

Though as you can probably guess, these do swap around from time to time
on kernel upgrades, so way better to use uuid or partition labels...

live well,
 vagrant

HTH
--
Ashish SHUKLA | GPG: F682 CDCC 39DC 0FEA E116  20B6 C746 CFA9 E74F A4B0
             | GPG: 01DE 145E 35D8 C87E 956E  FEC9 D4C4 4BDA 2C98 C654

"If I destroy you, what business is it of yours ?" (Dark Forest, Liu Cixin)

signature.asc
Description: PGP signature