If I cleanup the buffer / cache if it reaches 10GB … the rock5b will survive … he’s not freezing …
so maybe my memory is faulty or something else.
I will contact the supplier and ask for a new one …
Maybe others with this issue can try to reproduce it ?!
and give a feedback
Try running memtestfor some time,
Try same thing with additional cooling, About year ago I had such issue with my nuc box, I added more RAM and it started to work unreliably. I found out that two RAM modules will heat up and then they are not that stable, same thing with additional fan solved that issue.
See:
This may be the same problem, where we found the solution.
So be sure to update the SPL bootloader on SPI-NOR, even if you are booting from another medium.
as I can reproduce this issue with
„memtester 15000M 3“
with all tested OS‘ , I think it‘s hardware related… and I was able to catch this
any further ideas?
is there a bootable memtest86 für rock5b out there, everything I tried failed booting from sdcard and usbstick…
Have you even tried flashing spi-nor?
in regard to your howto? yes I did, but after that I wasn‘t able to boot anything. I had to recover the boot sector. so i guess I did something wrong….
I‘m not sure if I did everything right when running through your howto.
maybe you can provide it in a personal message or mail in german.?
Check your pms 
this solved my problem,
As I was in contact with radxa support, I told them they should have a look into this post to provide the SPI on their homepage.
Big Props to the guys who found this solution at write the howto for the medium skilled once 
same issue here: in my case I’m just performing an apt upgrade with official image, to be more precise a “sudo dpkg --configure -a” after an interrupted “apt upgrade”. I also tried to launch these commands inside a screen session, when I go back via ssh it says “no screen to resume”.
uname -a
Linux apollo 5.10.110-15-rockchip #a2d3d7e5d SMP Mon Sep 4 11:55:03 UTC 2023 aarch64 GNU/Linux
at the end of dmesg I have this, I don’t know if it is enough and I don’t know if bluetooth may cause this since it is connected via ethernet cable:
[ 7.581783] rk-pcie fe170000.pcie: PCIe Link Fail
[ 7.581802] rk-pcie fe170000.pcie: failed to initialize host
[ 7.608441] rk-pcie fe150000.pcie: PCIe Link Fail
[ 7.608458] rk-pcie fe150000.pcie: failed to initialize host
[ 7.608703] rockchip-pm-domain fd8d8000.power-management:power-controller: Looking up pcie-supply from device tree
[ 7.608721] rockchip-pm-domain fd8d8000.power-management:power-controller: Looking up pcie-supply property in node /power-management@fd8d8000/power-controller failed
[ 7.751716] mali fb000000.gpu: Loading Mali firmware 0x1010000
[ 7.752381] mali fb000000.gpu: Mali firmware git_sha: 9b8db9aa05a7b4f3066fd75dbdff695262f891fe
[ 7.902214] ttyFIQ ttyFIQ0: tty_port_close_start: tty->count = 1 port count = 2
[ 9.225174] Bluetooth: hci0: command 0xfc18 tx timeout
[ 9.729717] r8125: enP4p65s0: link up
[ 9.729746] IPv6: ADDRCONF(NETDEV_CHANGE): enP4p65s0: link becomes ready
[ 17.328701] Bluetooth: hci0: BCM: failed to write update baudrate (-110)
[ 17.328742] Bluetooth: hci0: Failed to set baudrate
[ 19.462295] Bluetooth: hci0: command 0x0c03 tx timeout
[ 27.355287] Bluetooth: hci0: BCM: Reset failed (-110)
I tried to reduce the memory frequency but no way. I tried to use only official parts and os to avoid compatibility issues, I wouldn’t expect so much instability.
If you need further information let me know.
UPDATE: I saw why screen or nohup launched dpkg won’t survive, I connected my rock5 and I run the same command directly from radxa desktop, at some point the device gets rebooted and the login screen appears again.
So it does look like the same problem. Did you update SPL loader on spi?