There is no need to save those u-boot sectors since the original copy is already located in /usr/lib/linux-u-boot-dev-orangepi3_20.08.0-trunk_arm64/ for example.
Instead of doing individual "cat /sys/bus/w1/devices/28-000003ebdd28/temperature" on each devices, I presume you can do multiple with "cat /sys/bus/w1/devices/*/temperature" ...
Oh ! I've completely forgot about the hack, I've forgot to add it in /etc/rc.local of the 5.8.y build ... I will check now ...
EDIT: Thanks for the reminder : it is not freezing any more, at least until now !
If you mean this patch/kernel/sunxi-dev/add-missing-H6-gpu-opp-in-5.7.y.patch, Yes it is still applied, although I had to remove CPU OPP since it became a duplicate.
But it most probably related to those ones since I had to disabled them completely (I could not easily fix them) :
patch/kernel/sunxi-dev/sun50i-h6-drm_panfrost-1-missing-remove-opp-table-in-case-of-failure.patch
patch/kernel/sunxi-dev/sun50i-h6-drm_panfrost-2-add-devfreq-regulator-support.patch
I've started the task ... But not all my Allwinner garden is tested ...
Yes ! I've done a patch to tweak sha256_state structure clashes for tons of Realtek drivers as well as API changes to cfg80211.
BTW, I've done some tests with BullEyes desktop on my OPi3, and Panfrost is freezing easily while it was working fine under 5.7.y.
You should change "verbosity" to "7" in /boot/armbianEnv.txt to be able to see more details after "Starting kernel".
EDIT: Sorry, I didn't saw your complete log URL ...
This version is far from been the newest, and I doubt it is supporting the "-@" option for symbols.
Download at least the 1.4.7 from 2018, from http://ftp.debian.org/debian/pool/main/d/device-tree-compiler/
I didn't say to short to GND permanently, but simply to check that the input is working fine as normal GPIO ...
It is already disabled by default, unless you load an I2C DT overlay ...
Pretty normal : PA19 is shared also as TWI1-SDA which is an I2C bus provided on header, and there are some PullUps resistors already present on the OPiZero board.
So, simply short PA19 to GND, you will see that your script will print "0".
Looking at my OPiOne+, it has different addresses for "script_addr" and "load_addr" and during u-boot init it shows :
## Executing script at 4fc00000
...
## Executing script at 44000000
EDIT : ... and here is what is shown for OrangePi-RK3399 :
## Executing script at 00500000
...
## Executing script at 09000000
That is maybe specific to EspressoBin, because on my Rockchip and Allwinner boards, it is as the following :
if test -e ${devtype} ${devnum} ${prefix}armbianEnv.txt; then
load ${devtype} ${devnum} ${load_addr} ${prefix}armbianEnv.txt
env import -t ${load_addr} ${filesize}
fi
Where "load_addr" varies depend of the SoC ...
What I found few months ago is that when using "old-school" /etc/network/interfaces, you need to do "systemctl disabled NetworkManager", otherwise latest is conflicting and try to steal interface.
Edit /boot/boot.cmd instead to remove the "console=ttyS0,115200" part, and recompile the /boot/boot.scr using "mkimage -C none -A arm -T script -d /boot/boot.cmd /boot/boot.scr".
Make sure /boot/armbianEnv.txt has "console=display" ...
The best option to have an IR Transmitter would be to use something like a BluePill STM32F103 board connected to the NanoPi, and having both IR RX and TX managed by this board, with proper firmware to also been able to "learn" IR codes that need to be emulated...