GPD Pocket


Yeah, looks like it's got its knickers in a twist during some typec_fusb302 initialisation stuff, judging by what comes just before that crash and the stack trace. It seems to have loaded and address from 00a0 (probably in the address space of the kernel object, mapped via a mmu, so that's not too crazy, on consideration) and getting 0 then trying to jump there (which is where it all falls down). It does then carry on detecting new filesystems that come online, but if that's stopped it initiing properly and creating a login for you then that's rather irrelevant.
 
I'm thinking it is a kernel problem now (updated my prev. post). Also thinking that having to power off with the power button once it hangs puts it into some sort of "weird" state and that is what all the blank screens and hanging is all about. So I'm guessing I'll have to wait for a newer kernel.

ljones
 
Yeah, there's 4.20 already out for a month or two now, and 5.0 waiting in the wings. Whether either of those will help you (or if indeed it's not actually a problem with the kernel itself but the modules it ships with) I couldn't say, plus 4.19 is a LTS kernel due to live until the end of 2020 so if the debian project decides to jump to anything newer before then I couldn't say. But in theory 4.19 should still carry any bugs found in later kernels and backported, so it might get better anyway.

Edit: Also, it thinking you need to fsck your filesystem suggests it's not always getting into a proper halt state before it hangs.
 
Is there currently any self contained x86 system that will run a fully supported Debian instance without hassles?
 
Yeah, there's 4.20 already out for a month or two now, and 5.0 waiting in the wings. Whether either of those will help you (or if indeed it's not actually a problem with the kernel itself but the modules it ships with) I couldn't say, plus 4.19 is a LTS kernel due to live until the end of 2020 so if the debian project decides to jump to anything newer before then I couldn't say. But in theory 4.19 should still carry any bugs found in later kernels and backported, so it might get better anyway.

Edit: Also, it thinking you need to fsck your filesystem suggests it's not always getting into a proper halt state before it hangs.

Ok, I found this site. First thing if I can even get it to boot properly will be to try to flash a newer bios from that site. Annoyingly the "2017/06/28 BIOS for Ubuntu" link is dead. I wonder what the bios for ubuntu even was?

Update: First thing I tried is to go back to kernel 4.9/Debian stretch and try that. But I couldn't get any display *at all* until I put in nomodeset fbcon=rotate:1 into the grub menu at boot (e at grub boot menu, change the option and then f10 to boot). Flashing seemed to work with flashrom as well. Now back to trying the newer debian/kernel 4.19.

I tried reinstalling 4.19/debian buster but still get the kernel crashes, though at least I have a screen without having to adjust grub.

Update: Managed to find that "bios for ubuntu" and it does seem to have some more options. Flashing was successful although I now need to try to reinstall 4.19/buster again. Still get the kernel crash however maybe there is some sort of bios option that needs to be disabled (unknown).

ljones
 
Last edited:
I'm not sure, but it looks like my "hard disk" (= flash) is broken after many installs (Debian, of course).
Does anybody know how to change it?
Many thanks in advance!
Btw. I bought the GPD Pocket (the first one, not "2") from Dragonbox, that's why I ask here.
 
Unfortunately, it can't be changed, it's soldered onto the PCB - but we can replace the full unit in this case.
 
Unfortunately, it can't be changed, it's soldered onto the PCB - but we can replace the full unit in this case.

Is there a way how I can test the flash? I don't want to pester you with this issue, if in the end it's another problem.

(Btw. I have had so many bad experience with all kind of flash memory in the past years, that I think soldering it onto the mainboard is a bad idea. Even "industrial" flash memory, i.e. SLC! with extended temperature range, 20 Euro for 2 GB, just broke more than once.)
 
Is there a way how I can test the flash? I don't want to pester you with this issue, if in the end it's another problem.

(Btw. I have had so many bad experience with all kind of flash memory in the past years, that I think soldering it onto the mainboard is a bad idea. Even "industrial" flash memory, i.e. SLC! with extended temperature range, 20 Euro for 2 GB, just broke more than once.)

What are you doing to all that poor little flash cells, you monster! ;)
Just read out the smart values, you can normally see the "wearing" level of the SSD.
In smarttools, the higher, the healthier.

sudo apt install smartmontools
check the ID of your SSD with lsblk
then
sudo smartctl -i /dev/sdX
for information.
Then the short test with:
sudo smartctl -t short -a /dev/sdX
if this already gives you errors -> send to ED
if not:
sudo smartctl -t long -a /dev/sdX
wich tests the whole drive reads and checks the"blocks".

NVM SSDs are names NVM0X or stuff like that, not sdX.
 
Hmmm, turned mine on today, and it reported low battery, so I shut it down and plugged it in; it's been sat on 10% charge for a couple of hours now. Not really sure what to do with it, as it seems to have been running okay when I've been using it.
 
Mine has a dead battery and I'm not sure where to get a new one. It bloated up massively while it was reporting all sorts of crazy battery levels.
 
Did you guys think of contacting company you bought your Pocket from?

That's where I'd start.....

Just a thought.....
 
Yeah, that's the problem with places like Gearbest for me, it's not like you can take it back with a receipt; and I don't imagine you get much in the way of a "warranty period", though I haven't checked.
Good call on ED having batteries in stock, not even touched my Pocket since, so need to get it open and just check everything first (might be a loose connection or something)
 
Back
Top