[Image] Rock Pi S v1.1 Slarm64 (aarch64 unofficial slackware)

With a different image (some armbian automated build), I get a serial console with “screen -L /dev/ttyUSB0 1500000”. But that one flashes no LEDs, eth stays dark, and complains about a missing devicetree file.

This one has no serial console whatsoever, tried with “screen -L /dev/ttyUSB0 1500000”, “screen -L /dev/ttyUSB0 115200” and “screen -L /dev/ttyUSB0 115200,cs8”. However, it once left a few lines in the dhcp server log, telling me hostname, MAC and IP address. It also flashes a blue LED and eth’s LEDs are on. After this, it just disappeared, no ping, no ssh.

I tried booting it more often, also tried to configure things in the microSD card, again a blue LED flashes and eth’s LEDs are on, but no more DHCP lines, and never any serial console.

Logs have nothing useful, there is only ever 1 boot trace even after booting many times, with 1970 date. Maybe it never booted this SD card at all.

What could be going on?

mara · April 8, 2021, 3:45pm

the image uses uart2 at a speed of 115200
pins:
RX(pin 21)| <—> TXD
TX(pin 19)| <—> RXD
GND(pin 20)| <—> GND

if the WIFI network the driver has an error.
otherwise it works.

mara · April 9, 2021, 4:08pm

update kernel 5.11.12
slarm64-current-aarch64-base-rock_pi_s-5.11.12-build-20210408.img.zst
slarm64-current-aarch64-base-rock_pi_s-5.11.12-build-20210408.img.zst.sha256

rockberry · April 11, 2021, 8:03pm

Thanks a lot, it works now with uart2. After creating a user account, ssh starts up.

mara · May 3, 2021, 8:09am

update kernel 5.12.1
slarm64-current-aarch64-base-rock_pi_s-5.12.1-build-20210502.img.zst
slarm64-current-aarch64-base-rock_pi_s-5.12.1-build-20210502.img.zst.sha256

rockberry · June 7, 2021, 10:19am

May I report something here?

I saw in /boot/boot.cmd these lines

if test “${console}” = “display” || test “${console}” = “both”; then setenv consoleargs “console=ttyS2,115200n8”; fi
if test “${console}” = “serial” || test “${console}” = “both”; then setenv consoleargs “${consoleargs} console=tty1”; fi

if display then use ttyS2
if serial then use tty1

Shouldn’t it be the other way?

So now I have one running with kernel 4.4 as a web proxy for ipv6, and one I’m testing with kernel 5.10. This crashes after a few days, and then never comes back up. Note, even with a serial console, there is nothing after this line: “RTL8201F Fast Ethernet stmmac-0:00: attached PHY driver [RTL8201F Fast Ethernet]” Basically, I must take out the SD card and fsck it. My Gentoo boxes do this automatically on unclean reboot, maybe it’s unsafe in theory.

I can only guess that something is running, but only on ttyS1, not ttyS2, and it has to do with fsck.

So I changed /etc/rc.d/rc.S and looked at the part “/sbin/fsck $FORCEFSCK -C -a /”. After this, if RETVAL -ge 4, it proceeds to sulogin. So, for next time, I commented out the PS1 line and the sulogin line, and added

/sbin/fsck -C -y /

It is ext4. (Maybe f2fs would work, but I never tried booting from it.) Now I hope it will work like Gentoo next time, try to auto-fix the filesystem and reboot.

rockberry · June 7, 2021, 10:32am

I found that changing the kernel for next boot is tricky. First, installing the 4.4 package unpacks into /boot/dtb and overwrites /boot/dtb-5.X.X, if such a kernel is active. The newer kernel packages install into their own directories.

After installing 4.4, one must fix the mess in /boot and then reinstall the previous 5.X kernel (if desired).

To activate a kernel, these symlinks must be set: uInitrd (make sure it is there), dtb, Image (Note!), vmlinux (possibly optional), config (optional) and System.map (optional)

I use this script /boot/set_kernel.sh (too lazy to deal with config and System.map).

#!/bin/sh

k=$1

[ “$k” ] || exit 1
#cd /boot
[ -r vmlinuz-"$k" ] || exit 2
[ -r uInitrd-"$k" ] || exit 3
[ -d dtb-"$k" ] || exit 3

ln -sf vmlinuz-"$k" vmlinuz
ln -sf vmlinuz-"$k" Image
ln -sf uInitrd-"$k" uInitrd
rm dtb
ln -s dtb-"$k" dtb

#Sorry this “indent preformatted text…” just emits the aforementioned text, there is no [code] tag

mara · June 7, 2021, 2:48pm

it looks like you are not using a package manager.

rockberry · June 9, 2021, 9:08am

How so? What am I doing wrong?

Note, the 3 kernels were installed with pkgtool. However, yes, usually I prefer to compile/install kernels myself. That’s the traditional Gentoo way, that’s how I can make some hardware work, and that would work around problems with armbian/raspbian (not pkgtool) removing the currently running kernel’s modules.

mara · June 9, 2021, 5:05pm

when using upgradepkg, the kernel is updated correctly