13:35 <naoki> Kwiboo: Strangely I got kernel hang during boot with ramboot or dfu in spl on ROCK 5B with 4GB RAM, but it doesn't happen on ROCK 5A 4GB

13:36 <naoki> Which device and rootfs media are you using for testing ramboot? Can you reach login: prompt?

13:37 <naoki> (I'll try again tomorrow...)

13:44 <naoki> Btw I can see some delay after usb472.bin sent on RK3399 and RK3328

13:46 <naoki> Interestingly, on RK3399 USB transfer is complete before SPL (and U-Boot) is booted, but on RK3328 USB transfer is not complete til SPL is booted

13:52 <naoki> just FYI... https://github.com/RadxaNaoki/rkramboot (another dirty hack..)

13:53 <naoki> I should learn rust...

13:59 <jakllsch> it's over-hyped

21:39 <Kwiboo> naoki: strange regarding only issue on 5B and not 5A, I am back close to my boards now so can do more runtime tests and prepare a ramboot RFC series

21:41 <Kwiboo> nice small tool :-), looking at https://github.com/xboot/xrock it seem to use a 10 ms delay after 471 before sending 472 command, and boot_merger encode the delay to use in the loader image

21:41 <naoki> I'll check other boards too, some of which just check that U-Boot starts.

21:43 <naoki> btw I noticed rkramboot should be rkusbboot ;)

21:43 <naoki> and I didn't know xrock... I need to check it

21:53 <Kwiboo> I have not used xrock myself but found it some time ago, could be useful, have not looked too close

22:01 <Kwiboo> for fast Linux testing I typically start a arm64 defconfig kernel (with built-in modules) with a minimal busybox rootfs, for more advanced testing I use a full ubuntu rootfs, just pushed bare minimum repo with scripts I use to build u-boot and generate a small boot.img.gz I can write to a sd-card

22:02 <naoki> from LICENSE file, xrock was written from 2007, but I guess it's not correct ;) anyway it looks good

22:02 <Kwiboo> https://github.com/Kwiboo/scripts/blob/main/minsys/docker-build.sh has some commands I used to build the minimal rootfs, for u-boot I just use e.g. ./docker-build.sh rock5b-rk3588