An unofficial Debian Installer for Pinebook Pro

[ab1jx]

Got it, thanks to p.H. on the Debian forums:

Code:

apt-get install exfat-fuse exfat-utils
cd /sbin
ln -s mount.exfat-fuse mount.exfat

May need a reboot, but then it works just like exfat in Stretch and Buster, both with mount and fsck.

[newton688]

Thanks All.

I finally gave this a try with an SD card and it worked fine after patching the script to remove network manager. The remainder of the Gnome/Wayland install seems to have resolved the NetworkManager dependencies. PR is here: https://github.com/daniel-thompson/pineb...er/pull/35

A couple of things I'm curious about. How hard would it be to get a newer kernel? I noticed that there are some things about audio controls that seem to be working a bit better with the Manjaro 5.11 kernel.

Also, how far are we at this point from being able to use the straight Debian installer?

[ab1jx]

Does anybody know how to turn this off?  My dmesg is cluttered with this junk

Code:

[254790.184071] Purging 8687616 bytes
[254794.238139] Purging 614400 bytes
[254794.295878] Purging 8548352 bytes
[254814.595327] Purging 561152 bytes
[254815.983201] Purging 532480 bytes
[254832.402633] Purging 528384 bytes
[254840.965064] Purging 557056 bytes
[254846.183914] Purging 524288 bytes
[254846.247936] Purging 565248 bytes
[254848.747086] Purging 622592 bytes
[254848.773149] Purging 540672 bytes
[254850.216978] Purging 565248 bytes
[254850.938291] Purging 561152 bytes
[254852.756999] Purging 8421376 bytes
[254857.352373] Purging 8646656 bytes
[254858.969830] Purging 720896 bytes
[254859.009549] Purging 634880 bytes
[254866.755186] Purging 8572928 bytes
[254873.558351] Purging 8417280 bytes
[254873.812790] Purging 581632 bytes
[254873.850428] Purging 8798208 bytes
[254882.134874] Purging 565248 bytes
[254882.203893] Purging 8433664 bytes
[254882.246202] Purging 8425472 bytes
[254882.304568] Purging 536576 bytes
[254882.518902] Purging 524288 bytes
[254882.540537] Purging 622592 bytes
[254882.672026] Purging 8699904 bytes
[254882.944469] Purging 528384 bytes
[254883.386020] Purging 8658944 bytes
[254883.635271] Purging 8519680 bytes

I just saw an uninterrupted 900 lines of it. Does it come from panfrost? Other than that it looks pretty good.

[newton688]

It's happened a couple of times that one of the filesystems remounts as read-only with something due to the mmc driver. I'll need to capture the messages somehow and post them. Anybody else seeing this?

I wonder if it's related to suspend.

[Der Geist der Maschine]

Hi all,

I'm on Debian 10. Is Debian 11/Bullseye stable?

Does Debian 11/Bullseye came with its own linux kernel deb? Is it usable or are folks still using Daniel's kernel repo?

[ab1jx]

I'm using Daniel's kernel, the only problem I've seen is that filesystems have been moved into the kernel since then so I had to add a symlink to make it work transparently. This helps with both mount and fsck, some equivalent may with other filesystems. I had posted a question on the Debian forums, but this isn't a Debian kernel so they wanted to know where I got it.

In /sbin,
ln -s mount.exfat-fuse mount.exfat

[KiteX3]

Does anybody know how to turn this off?  My dmesg is cluttered with this junk

Code:

[254790.184071] Purging 8687616 bytes
[254794.238139] Purging 614400 bytes
[254794.295878] Purging 8548352 bytes
[254814.595327] Purging 561152 bytes
[254815.983201] Purging 532480 bytes
[254832.402633] Purging 528384 bytes
[254840.965064] Purging 557056 bytes
[254846.183914] Purging 524288 bytes
[254846.247936] Purging 565248 bytes
[254848.747086] Purging 622592 bytes
[254848.773149] Purging 540672 bytes
[254850.216978] Purging 565248 bytes
[254850.938291] Purging 561152 bytes
[254852.756999] Purging 8421376 bytes
[254857.352373] Purging 8646656 bytes
[254858.969830] Purging 720896 bytes
[254859.009549] Purging 634880 bytes
[254866.755186] Purging 8572928 bytes
[254873.558351] Purging 8417280 bytes
[254873.812790] Purging 581632 bytes
[254873.850428] Purging 8798208 bytes
[254882.134874] Purging 565248 bytes
[254882.203893] Purging 8433664 bytes
[254882.246202] Purging 8425472 bytes
[254882.304568] Purging 536576 bytes
[254882.518902] Purging 524288 bytes
[254882.540537] Purging 622592 bytes
[254882.672026] Purging 8699904 bytes
[254882.944469] Purging 528384 bytes
[254883.386020] Purging 8658944 bytes
[254883.635271] Purging 8519680 bytes

I just saw an uninterrupted  900 lines of it.  Does it come from panfrost?  Other than that it looks pretty good.

I have been wondering about something related to this too; I'm getting these, but also in particular I'm getting reproduce-able crashes while either using GIMP for any period of time or visiting certain webpages in Chromium with panfrost error messages clogging up my dmesg -wH. Both the internal display and my external monitor freeze at that point with errors saying:

Code:

panfrost ff9a0000.gpu: Unhandled Page fault in AS1 at VA 0x(lots of numbers that vary)
Reason: TODO
raw fault status: (some hexidecimal code that varies)
decoded fault status: SLAVE FAULT
exception type 0xC1: TRANSLATION_FAULT_LEVEL1
access type 0x2: READ
source id (hexidecimal code that varies)

After this, the screens freeze entirely, but audio continues working just fine. Even the good old Ctrl+Alt+F(N) virtual terminal trick doesn't work to get the computer working; I have to force reboot to get that state to stop.

Anyone experiencing the same issues? Anyone have any idea how to fix this?

[ab1jx]

It's just a video freeze-up I think, connecting by ssh to shut it down or reboot still works.  I've become a fan or Termux on phones for that reason.  Just pull out a phone and use it to reboot your PBP, mostly more convenient than walking to another computer. It's crashing the video at a level lower than X, that's why you can't switch to another virtual teminal.

In my case Panfrost wasn't causing the crashes, they were because the original eMMC was starting to fail so something would try to read or write on a bad spot.  My workaround was to set up the Daniel Thompson installation on a new sd card and use it to boot from.  I get uptimes of a week or so, it's quite stable.  My original eMMC is still in place as a backup boot medium.  I have an nvme drive in here but not booting yet, so I periodically boot from the eMMC then use dd to backup my sd card to an image on the nvme.

Actually that doesn't make a lot of sense, a crash due to the eMMC might leave it without ssh and totally dead. But it hasn't happened in months, I was trying to remember exactly. But it was the eMMC because that's the only thing I changed.

I suppose you could find a Panfrost forum or mailing list and ask there about the messages.  I expected them to be cured by the apt-get update/upgrade cycle but it's been over a year now. This Panfrost is considerably slower than the video in the mrfixit Stretch. And there's a bug with having your speakers stay on after you plug in headphones that wasn't in the mrfixit image.

Look at the grep man page, it's possible to set it to filter out lines containing a phrase to clean up your dmesg, then pipe it to something else.

[KiteX3]

It seems that it still hasn't been fixed; I'm still experiencing GIMP crashes of that type. Unusually, the mouse cursor still seems to be movable even while the rest of the display (GIMP, other apps, windows, and KDE applets like the clock) is frozen, which seems to contradict the idea that it's crashing at a level below that of X; if so, wouldn't it freeze the cursor too? (It does freeze the cursor if I hit Ctrl+Alt+F2 to get to a virtual terminal though.)

Oddly, I don't recall ever encountering these problems on Manjaro; it had its own assortment of problems, but I recall GIMP functioning just fine.

A similar bug is also occurring with Chromium on occasion, especially on stupidly over-complicated websites.

[ab1jx]

I've never noticed a problem with Gimp but then I don't try to run it with KDE.  In LXDE the clock updates so infrequently it's not worth using it, sometimes it's 20 minutes or more behind what you get if you just type date into  a terminal emulator.  Because I use startx I guess instead of lightdm I can ctrl-alt-f2 and I'm at a different virtual console, but I've never found Wayland useful for anything.

I think the LXDE clock just runs at too low a priority so it doesn't get updated often enough.  xclock works.  I use only Alsa sound and sometimes that's busy.  But doing a pkill of Firefox (I never use Chromium) will free up the sound, then letting Firefox restore the session works.  If I close it with the normal menu I lose all my tabs.

Sometimes I miss FVWM for its simplicity but I don't think I've gotten it to work under Linux at all.  It's the default in OpenBSD last I knew.