[SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Discussion about U-Boot and the kernel.

Re: [SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Postby Kabbone » Sat Aug 22, 2020 8:10 am

mbakhterev wrote:ANOTHER WARNING: The installer of the uboot-rock64-2020.07 package unconditionally flashes images to the /dev/mmcblk0, which was in my case my encrypted SD with home directory. I just did not pay enough attention to the message (previous version of uboot-rock64 was detecting name of boot EMMC correctly) and i've irrecoverably f*cked up some of my data. BE CAREFUL!


There was no change to the install script in the update, it's the same as before. EMMC was never detected, it was hardcoded to mmcblk0
Kabbone
 
Posts: 137
Joined: Thu Jul 25, 2013 9:20 am

Re: [SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Postby mbakhterev » Sat Aug 22, 2020 10:33 am

Kabbone wrote:
There was no change to the install script in the update, it's the same as before. EMMC was never detected, it was hardcoded to mmcblk0


Indeed! On my previous setup sensitive data was on usb drive, and i had no sd card. So, as i figured out now, install script was just happily creating plain /dev/mmcblk0 file, and bootloader was never updated since install, while i was pretty sure, that everything was fine.

Should the install script contain more checks, or should the message about target device be more explictit? But, personally, i am against complex solutions. It is better to include explicit install script in the package, so user could call something like "/boot/flash-uboot device" after double and triple checking.
mbakhterev
 
Posts: 20
Joined: Mon Oct 14, 2019 4:45 pm

Re: [SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Postby mbakhterev » Tue Aug 25, 2020 6:25 pm

After i started to run medium computation loads (some network services and compilation), memory instability showed up. What is careful way to decrease memory frequency? Is replacing 800 with 768 in rk3328-sdram-lpddr3-1600.dtsi safe? Or is it better to replace rk3328-sdram-lpddr3-1600.dts with rk3328-sdram-lpddr3-666.dts in rk3328-rock64-u-boot.dtsi?
mbakhterev
 
Posts: 20
Joined: Mon Oct 14, 2019 4:45 pm

Re: [SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Postby Kabbone » Tue Aug 25, 2020 6:39 pm

mbakhterev wrote:After i started to run medium computation loads (some network services and compilation), memory instability showed up. What is careful way to decrease memory frequency? Is replacing 800 with 768 in rk3328-sdram-lpddr3-1600.dtsi safe? Or is it better to replace rk3328-sdram-lpddr3-1600.dts with rk3328-sdram-lpddr3-666.dts in rk3328-rock64-u-boot.dtsi?


You can just replace the 800 with the desired value, at least it works for me. I clock down as far as 600, but I already had problems before.
Kabbone
 
Posts: 137
Joined: Thu Jul 25, 2013 9:20 am

Re: [SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Postby mbakhterev » Thu Aug 27, 2020 3:02 pm

Kabbone wrote:You can just replace the 800 with the desired value, at least it works for me. I clock down as far as 600, but I already had problems before.


Thanks. 600 is working for me. I've tested with heavy html pages. long-lasting compilation workloads and youtube video playbacks.

But the HDMI output glitches intensively. Even with acceleration disabled. Looks like some kind of vertical desynchronization.
mbakhterev
 
Posts: 20
Joined: Mon Oct 14, 2019 4:45 pm

Re: [SOLVED] linux-aarch64-5.8.0-1 breaks Rock64

Postby ramsey » Wed Sep 09, 2020 11:47 pm

Shouldn't we update the PKGBUILD of uboot-rock64?


I considered it. What stops me is the requirement to cross compile ATF for arm32. I couldn't find the ALARM package to do that when I was building, so I went with a debian container which had the package. Also, I prefer blob free source, but I suspect someone might have the opposite preference. What to do in that situation? Two packages? source and bin?
ramsey
 
Posts: 25
Joined: Mon Aug 14, 2017 3:46 pm

Previous

Return to U-Boot/Kernel

Who is online

Users browsing this forum: No registered users and 1 guest