HP Omnibook 5 16" OLED - Purwa/x1p-42-100

[JeremiahCornelius]

I'm kicking off a discussion here, branching from the "Concept Snapdragon" forum on Ubuntu Discourse. Jens suggested I continue the side-topic on HP Omnibook 5 here.

I got a hold of an HP Omnibook 5 16” OLED with 1TB storage and 64 GB RAM. These are cheap and plentiful, well-made x1p-42-100 machines. I got this one for 800 USD, taxed and shipped.

It would be a real service to get these machines functional under a user-friendly Linux.

Confession: I'm an aarch64 junkie. Stop me before I buy another board for Armbian.

Since I have built system images with debbootstrap, a DTB folder and kernel-sources many, many times - and used these techniques to put Debian on Apple Silicon from the Asahi base, I assumed getting a baseline Linux on this machine would be easier. Haha!

Rather than re-hash the Discourse thread, I link the top message of our sub-topic here:

https://discourse.ubuntu.com/t/ubuntu-concept-snapdragon-x-elite/48800/1359

[JeremiahCornelius]

Short update on status.
I get regular, reliable USB boot with the image questing-desktop-arm64+x1e-20260119_extended_jg.iso from Jens' g-drive share, selecting the DTP for "HP Omnibook X" from the pre-built GRUB selections.
https://drive.google.com/drive/folders/1sc_CpqOMTJNljfvRyLG-xdwB0yduje_O

• The OLED renders perfectly, without my having tested for acceleration. Resolution and refresh rate are correct, and 125% fractional scaling makes a very useable desktop screeen.
• The subiquity installer is broken, and consistently fails.
• No audio. I haven't yet tried the Omnibook audio enabling fixes, published elsewhere.
• Trackpad HID doesn't work, with good input on the Discourse thread about the relatively simple fix in a device-specific DTB for this model.
• Battery monitoring and touchscreen HID are functional
• Function-key screen-brightess controls do nothing. The GNOME shell slider widget is present, but does nothing.
• No WiFi. Normally a big PITA, but I have a nice workaround.

Network problems are overcome by USB tether to an iPhone with Mobile Hotspot enabled. Out of the box, this Questing on the 6.19 kernel just works. Devices both produce dialogs for connection trust, and then the Gnome Shell network widget changes state.

ping yahoo.com Yes. We have Internet. I'm now able to browse to my Nextcloud instance, and persistent storage is available to ship configs from my functional lab machines, to the Omnibook 5 as I begin a manual build from the booted image.

[JeremiahCornelius]

This is my "light at the end of the tunnel" post.

It's suggested that many images can't make full use of available RAM, and I have 64GB on this example.
Testing was suggested, by transferring a very large file over USB. I mounted the NTFS partition and transferred hiberfil.sys from the Windows C: root to the image's tmpfs on /var/crash. That was 13 GB at roughly 8MBs.
Later, I'll set up a bigger file to test, but this worked as expected.

Second, I made a new BTRFS partition in 800GB freespace on the NVMe.
Next, I can setup subvols to support a @ root and @home, along with other subvols that manage quirks of Gnome/GDM etc, for clean snapshotting. I like building this way. I can swap out whole different root FS builds, with a change to GRUB and a reboot.

I'll report back, after I bust out subvols, rsync a bunch into a chroot, and start tinkering.

EXAMPLE:

/etc/fstab: static file system information.

 <file system>             <mount point>  <type>  <options>  <dump>  <pass>
UUID=8CA4-47C6                            /boot/efi      vfat    defaults,noatime 0 2
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /              btrfs   subvol=/@,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /.snapshots    btrfs   subvol=/@snapshots,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /home          btrfs   subvol=/@home,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /root          btrfs   subvol=/@root,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /var/log       btrfs   subvol=/@var@log,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /var/lib/AccountsService btrfs   subvol=/@var@lib@AccountsService,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /var/lib/gdm3  btrfs   subvol=/@var@lib@gdm3,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-c7bca982-cb96-48c7-82ad-766248a5aa42 /tmp           btrfs   subvol=/@tmp,defaults,noatime,space_cache=v2,compress=zstd 0 0
/dev/mapper/luks-db482100-c1b1-4bc4-ad7a-07a5cfce23a2 swap           swap    defaults,noatime 0 0
tmpfs                                     /tmp           tmpfs   defaults,noatime,mode=1777 0 0


btrfs subvolumes:

ID 256 gen 554465 top level 5 path @
ID 257 gen 470984 top level 5 path @snapshots
ID 258 gen 554464 top level 5 path @home
ID 259 gen 457980 top level 5 path @root
ID 260 gen 554465 top level 5 path @var@log
ID 261 gen 466267 top level 5 path @var@lib@AccountsService
ID 262 gen 466268 top level 5 path @var@lib@gdm3
ID 263 gen 554465 top level 5 path @tmp
ID 265 gen 554403 top level 258 path @home/.snapshots

Also build subvolume: path @var@lib@docker

[jglathe]

Hi @JeremiahCornelius , welcome here. For bringup its good to have these extracts (easiest to get from Windows):

• ACPI tables readout and disassembly for GPIO and config mining, like here
• Log from HWInfo64 which is surprisingly helpful to identify buses used and where devices should be
• AeoB extract from the Windows drivers like here. These are helpful for GPIO configs that are not in the ACPI tables.
  Also a good idea to keep an operational Windows on the SSD (I usually do). For firmware fetching and info extraction, and if you need to boot an OS known to work on the HW for

[JeremiahCornelius]

Thanks for inviting me and the welcome. I will get to the extraction of these details tomorrow, and supply these with the aim of contributing to an eventual device specific DTB, etc.

Yeah, I reserved 70GB for the Windows 11 partition. Even with a heavily configured WSL environment, it's a pain to use. I think that the windows terminals being so bad were a driving motivation for the creation of popular IDEs! LOL

If I were to build a debbootstrap resembling your January questing-desktop-arm64+x1e-20260119_extended_jg.iso then I'd include the four "no-stubble" packages from https://drive.google.com/drive/folders/1n3Ii_k7bKVV5gIUkoBruUz2DXkEqBVFS , right?

• linux-buildinfo-6.19.0-rc8-jg-1-qcom-x1e_6.19.0-rc8-jg-1_arm64.deb
• linux-headers-6.19.0-rc8-jg-1-qcom-x1e_6.19.0-rc8-jg-1_arm64.deb
• linux-image-6.19.0-rc8-jg-1-qcom-x1e_6.19.0-rc8-jg-1_arm64.deb
• linux-modules-6.19.0-rc8-jg-1-qcom-x1e_6.19.0-rc8-jg-1_arm64.deb

I assume these are the Ubuntu/Debian packages, and those labeled "qcom" are for the devkit hardware, but it like to validate this.

[jglathe]

Ubuntu Concept is using the name qcom-x1e throughout their repo: https://git.launchpad.net/~ubuntu-concept/ubuntu/+source/linux/+git/questing Since my tree mainly is based on the Ubuntu tree, I use the same naming. Also the same default config, plus a few small bits. So the kernel packages I build should be drop-in replacements, yes. The actual hardware supported is... all arm64 that Canonical chooses to support. Meaning also Rockchip, Broadcom, CIX. My repo (this one) also has a rebuild.sh borrowed from pop-os to build the whole packages as defined by Canonical.

[jglathe]

A dmesg output and lsusb -tv would probably be great.