An unofficial Debian Installer for Pinebook Pro

[wasgurd]

Code:

sudo update-initramfs -u
update-initramfs: Generating /boot/initrd.img-5.4.2-2-pinebookpro-arm64
find: ‘/lib/modules/5.4.2-2-pinebookpro-arm64/kernel/arch/*/crypto’: No such file or directory
depmod: ERROR: ../libkmod/libkmod.c:515 lookup_builtin_file() could not open builtin file '/var/tmp/mkinitramfs_4g46bd/lib/modules/5.4.2-2-pinebookpro-arm64/modules.builtin.bin'

Should I ignore it, in general?

[Jeremiah Cornelius]

Alright! I have Daniel's Bullsye with Gnome3 installed to my EMMc on the ANSI PBP, and I feel like I'm in front of a "real" computer now!
I'm going to migrate a bunch of my extensions from the regular setup I use on a MacbookPro running Pop_OS, and I'm almost set for a "stable" machine, with the 2nd hand ISO model I bought for a "hacker".

All that I really need now is improvements to this trackpad. Even with firmware updates, it's a bear. I'd do a 3rd party swap on the part, if one is ever spec'd.

[xmixahlx]

in order to install armhf system I removed the ARCH=arm64 line from the install script and passed ARCH=armhf to the install command

[ThatGeoGuy]

I'm trying to figure out what the rough edges would be, but not having encryption support on a laptop is a pretty big one for me.I assume by default most of the inputs work, and WiFi / Bluetooth is in good shape? If so, it doesn't seem too limited in comparison to default debian, so I'll take your word for it. My plan is to avoid most DEs anyways, and probably stick to awesomewm/sway/i3 anyways. So assuming the basics are working, I don't think my needs are too complex

WiFi works but I'm not using (nor testing) bluetooth at the moment. I have seen some people have had difficulty making it work (check out the mainline kernel thread for clues... IIRC there was something about manually supplying a MAC address).

To be clear I am using a distro built by (a very old version) of the installer as my only OS on the PBP. It's robust and I'm now using PBP to build it's own kernel and bootloaders.

Having said that I'm not spending quite as much time on PBP as I might like... because I have a  PineTime too ;-) .

Alas, it seems bluetooth doesn't work. Relevant thread on the forums: https://forum.pine64.org/showthread.php?...2#pid57292

Has anyone in here had any success with this?

[Jeremiah Cornelius]

in order to install armhf system I removed the ARCH=arm64 line from the install script and passed ARCH=armhf to the install command

This is great. I'm midway into an install that reproduces your configuration right now. Widevine and other RPi binaries should now be possible to execute.

I'm also a weirdo, and seem to be preferring the original trackpad firmware to the update for dragging, etc. Glad I have 2 PBPs for these experiments.

[Jeremiah Cornelius]

in order to install armhf system I removed the ARCH=arm64 line from the install script and passed ARCH=armhf to the install command

This is great. I'm midway into an install that reproduces your configuration right now. Widevine and other RPi binaries should now be possible to execute.

I'm also a weirdo, and seem to be preferring the original trackpad firmware to the update for dragging, etc. Glad I have 2 PBPs for these experiments.

Good news on this front! My armhf userland install works great on the ISO PBP, including encrypted LUKS. One of the problems of the pure aarch64 installation is a non-launching Firefox. This is solved matching the mrfixit-style dual arch. This means Wildvine will work with some pipefitting, too.

Incidentally, choosing a UK 105 KB with a US language and US UTF-8 locale at install gave me perfect keybindings, and the £ key and # key are on their marked locations. Good for any North Americans installing Debian on their ISO machines.

[xmixahlx]

@danielt script feedback:

first, tyvm!!! here are some ideas:

* change ARCH handling in script ("ARCH=arm64" hardcoded disabled armhf userspace install)
* recommend requiring BLKDEV and ARCH at commandline to confirm explicit request
* setup hostname in script
* setup root password in script
* use 5.5-rc5 tsys kernel for improvements

other debian topics seem to be:

* determine bluetooth fixes needed
* determine what breaks upgrading to debian unstable (xserver packages???)
* test panfrost performance?
* document DM, DE and xorg/wayland performance/status

[richbl]

Something has happened with firefox-esr in Debian bullseye, it fails to start with an 'illegal instruction' message

Enviado desde mi ONEPLUS A5010 mediante Tapatalk

I've gotten the same, to resolve I've installed the firefox-esr from the buster-backports. It works fine for me.

Running Debian Bullseye from this installer, and when the native Firefox install failed, I too installed the Buster backport (Mozilla Firefox 68.4.1esr), but still no joy, with Firefox failing with the same "illegal instruction" error. Tried running under both X11 and Wayland, with no change in error output.

I'm wondering if there's something else floating around that's causing the "illegal instruction" error. Maybe the original Firefox purge didn't clean up all of the libs?

I can say that the rendering problem in Epiphany (screen blanking/white screens) goes away under Wayland. 

BTW, spinning up Debian with Gnome has been a great win for my PBP (just received it on Friday). Thanks Daniel!

Rich

[xmixahlx]

you can install firefox and/or chromium from unstable, also.

[danielt]

Wow! Looks like its been a busy weekend.


For those people having trouble with dptx.bin I noticed that @NigelT has posted how to fix this. Also @xmixahlx has posted some feedback on ways to improve the script. I did fix the ARCH=armhf bug since that was pretty crass but for now I've left the rest alone... I will be hacking on my PineTime for a bit longer yet! However I will still be reviewing github pull requests so if anyone wants to look at further fixes and improvements then PRs are very welcome!

Also @amp asked about the status of the upstream Debian installer. AFAIK at the moment there are still too many pieces missing to get very far (PineBook bootloader is missing some useful features and I don't think the kernel patch to enable the display will arrive until v5.6). I spent sometime over Christmas working on the bootloader but I have been overtaken by events and will eventually take a fresh look.