Intermittent Boot Issues

[MrRoosevelt]

Greetings

My girlfriend and I both got our PBPs in last week. Out of the box I had issues with mine not booting properly.

I got the Manjaro setup process then on reboot, I got the orange to green led sequence but no back light on the screen. After a few power cycles I was able to boot into Manjaro. Every other power cycle the issue returns though. I can do a hard poweroff and power back up and it fixes the issue; though, it sometimes takes 3 or 4 power cycles to fix the issue. There seems to be no relation between power connectivity across boots as I have read in other posts on this forum. I have also tried switching ttys and trying to reboot through the serial console, as if the screen were just blanked, which was also fruitless.

I'm running the PBP as shipped from the factory; though I did allow Manjaro to do a system update once it finally booted.

My girlfriend's PBP, on the other hand, is behaving quite nicely. 

I'm not sure where to start with debugging this. It looks like a hardware issue given that its not happening on both PBPs; but, I don't want to rule out the possibility that the Manjaro setup script botched something. Does anyone have any ideas?

Thanks in advance.

[Solution]

The solution as mentioned by @khanku is that there are bugs in Manjaro's build of UBoot.

There is a uboot build in this git repo: https://github.com/mrfixit2001/updates_repo/tree/v2.0 The v2.0 branch targets PBP. There is a script in the directory pinebook/filesystem called mrfixit_update.sh. The script is smart enough to determine the target drive for writing the uboot image. I didn't want to install the kernel and other stuff, so I borrowed the script and made some redactions to fit my needs.

This may not be the best way to handle it; but, its what I did to get moving:

Code:

DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
SYSPART=$(findmnt -n -o SOURCE /)
       
if echo $SYSPART | grep -qE 'p[0-9]$' ; then
  DEVID=$(echo $SYSPART | sed -e s+'p[0-9]$'+''+)
else
  DEVID=$(echo $SYSPART | sed -e s+'[0-9]$'++)
fi
       
if [ -f $DIR/idbloader.img ] ; then
  dd if=$DIR/idbloader.img of=$DEVID bs=32k seek=1 conv=fsync
fi

if [ -f $DIR/uboot.img ] ; then
  dd if=$DIR/uboot.img of=$DEVID bs=64k seek=128 conv=fsync
fi
       
if [ -f $DIR/trust.img ] ; then
  dd if=$DIR/trust.img of=$DEVID bs=64k seek=192 conv=fsync
fi

I ran these commands one at a time; but, some return value checks on dd and better placement of file existence would help automation.

[KiteX3]

I have also been encountering almost exactly this same problem with my brand new Pinebook Pro; unfortunately I have no idea what's causing it.

[jiyong]

My first guess would be to open the PBP and check if the eMMC is properly seated.
Be careful with static electricity.

[khanku]

I got the Manjaro setup process then on reboot, I got the orange to green led sequence but no back light on the screen. After a few power cycles I was able to boot into Manjaro. Every other power cycle the issue returns though. I can do a hard poweroff and power back up and it fixes the issue; though, it sometimes takes 3 or 4 power cycles to fix the issue. There seems to be no relation between power connectivity across boots as I have read in other posts on this forum. I have also tried switching ttys and trying to reboot through the serial console, as if the screen were just blanked, which was also fruitless.

I had the same issue: Manjaro would boot every third of fourth try. I fixed this by writing u-boot files from mrfixit2001 updates_repo to the eMMC. While this worked for me I cannot tell you why and could not find the sources for those files

[Arwen]

I too have had serious Manjaro boot problems. Miy PBP was from an earlier batch, so I had loaded Manjaro myself. Seemed to work a bit. Then it would not. Initially I could do the Control-Alt-Delete and then it would boot.

But in the end, I think I have to try something different. Since it appears to be a U-Boot issue, I'll give @khanku fix a try.

[Twig]

Having similar issues. Manjaro crashed while copying some non essential files and has since been unable to boot. I can get the orange and sometimes green light but nothing on the screen. I guess I'll need to check the emmc?

[tophneal]

This might be worth bringing up to the Manjaro devs on their forums. Sounds like a bad update to U-Boot went out if it's resolved by writing a BSP U-Boot. (one the up side, BSP U-Boot makes suspend possible in Manjaro.)

[yurievitch]

I got the Manjaro setup process then on reboot, I got the orange to green led sequence but no back light on the screen. After a few power cycles I was able to boot into Manjaro. Every other power cycle the issue returns though. I can do a hard poweroff and power back up and it fixes the issue; though, it sometimes takes 3 or 4 power cycles to fix the issue. There seems to be no relation between power connectivity across boots as I have read in other posts on this forum. I have also tried switching ttys and trying to reboot through the serial console, as if the screen were just blanked, which was also fruitless.

I had the same issue: Manjaro would boot every third of fourth try. I fixed this by writing u-boot files from mrfixit2001 updates_repo to the eMMC. While this worked for me I cannot tell you why and could not find the sources for those files

Same thing here. Is there an F key that lets me choose what device to boot to? I want to install Ubuntu Mate via a usb thumb drive.

[Arwen]

The installation of the default Debian U-Boot helped.

I've updated the Manjaro packages, which included some new U-Boot files. Those seem to work for the moment.
But, I am keeping the default Debian U-Boot files just in case.

I wish I had more skills in software development. But, that is so last century for me :-).
If time & brain power permit, I may dig into U-Boot on our Pinebook Pros. I really want more modern features.

Edit: That did not last long, had to go back to the default Debian U-Boot files.

[MrRoosevelt]

I  get denied when cloning or downloading the mrfixit repo.

[EDIT]: Nevermind, that was user error...

I'm trying to make sure that I write the images to the correct device. It looks like the script wants to do this (omitting long file path):

Code:

dd if=...updates_repo/pinebook/filesystem/idbloader.img of=/dev/mmcblk2 bs=32k seek=1 conv=fsync &>/dev/null
dd if=...updates_repo/pinebook/filesystem/uboot.img of=/dev/mmcblk2 bs=64k seek=128 conv=fsync &>/dev/null
dd if=...updates_repo/pinebook/filesystem/trust.img of=/dev/mmcblk2 bs=64k seek=192 conv=fsync &>/dev/null

It looks sane to me; but, I'm still looking for documentation on the wiki to confirm (I don't want to overwrite the wrong device).