Acer Swift 14 Go appears not to boot

[Cpt-Dingus]

Hi, I'm glad to see this device being worked on! I've downloaded and attempted to run the image linked here, have finally gotten the image to boot! However, the image does not install, appears to hang on Configuring keyboard... with the logs not existing because of tail: can't open /var/log/installer/debugand tail: can't open /var/log/syslog because neither exist.

I also want to point out that my first boot took a few minutes to get to the installer screen, the second it got to it the device went back to the kernel which just said Terminated. without any errors. On my second boot it went through the installer near instantly.

[Cpt-Dingus]

Just porting the update from the ubuntu discussion thread, it booted fine upon a reflash! Seems to work fine for the most part, can do more testing when I am able to move the install to my NVME.

[jglathe]

Now that it is running (it's not really an installer, its pre-installed desktop like on a Raspberry Pi), let's see if you can get it onto the nvme.

Run it on the nvme: You need some courage, and maybe a recovery stick if it goes south (no idea what Acer has to offer in this regard). Make sure you have a backup of all your data on this laptop.

• 1TB drive: shrink Windows root partition to half (that's what I do normally). Gives you ~500GB space for Linux.
• have the linux boot image itself available either as a second stick or on the stick itself (needs to be 32GB at least for that, needs to be decompressed as .img)
• boot from the stick
• install dosfstools mtools
• run gparted, add the root partition for your new system, give a distinct label (NOT the same as on the stick), format it the way you prefer (ext4 is default and works well, I have one box with btrfs, works too. Mount scripts and fstab differ, though. I paid attention that the usual filesystems are built-in in the kernel).
• mount the new partition directly (assuming it is /dev/nvme0n1p5, ESP is /dev/nvme0n1p1)
  sudo mount /dev/nvme0n1p5 /mnt/
• mount the linux boot image (simply with the disks tool, your path should be /media/<your username>/root2410
• rsync the whole freaking rootfs over to the new partition
  sudo rsync -avE /media/<your username>/root2410/ /mnt/

Now comes the more complicated part. We have the rootfs installed (which will do its "fetch firmware / reboot / set up user" dance when used), but we need to:

• install the grub bootloader to the nvme
• make a boot entry for this system
• move over to boot from the nvme without destroying anything.

To achieve this we need to do this in several reboot steps:

1. Prepare to boot from the stick, but with the rootfs on the nvme
2. Rewire for nvme boot
3. Boot from nvme

1. Prepare to boot from the stick, but with the rootfs on the nvme

For this we need to edit /mnt/etc/fstab for the first boot.

sudo nano /mnt/etc/fstab

Actually, this is two changes - we define the local nvme EFI partition to be mounted on /boot/efi/, and our new root filesystem on /. This will also be the final config for the new system.

To make a boot entry for this system, we use update-grub. We need to amend the generated file for booting with a device tree, though.:

# add boot entries from other partitions
sudo update-grub

# edit the boot entry after generating - unfortunaly necessary due to missing functionality (don't want to say bug) in grub
nano /boot/grub/grub.cfg

# add the the device-tree  /boot/dtb line to the newly created entry pointing to /dev/nvme0n1. You can copy the line from the first entry and add it to all Linux entries you can find where it is missing (should be 2)

	devicetree	/boot/dtb-6.15.0-rc1-jg-3-qcom-x1e

After that, we can reboot from stick, with the newly created boot entry from nvme0n1. It should come up (after its setup dance with one reboot).

2. Rewire for nvme boot

We're running now on the rootfs that is already on the nvme. We have already mounted the EFI partition on the nvme, too. The remaining steps are to install grub-efi and update grub again.

# install grub - this adds the uefi boot entry, too. 
sudo grub-install

#normally we should be able to remove the USB stick we booted from - makes for a cleaner grub boot menu
sudo update-grub

Now it should be possible to reboot from nvme, and dual-boot

3. Boot from nvme

The reboot without the USB stick should bring up grub and boot the system from nvme, or Windows, depending on what you choose. If it just boots windows (not unusual, unfortunately) then we need to change the boot order in the UEFI BIOS to have Ubuntu first.

[Cpt-Dingus]

The issue is that grub doesn't seemingly even show up at all, how would linux options affect it? I can't even get the selection screen to pop up.

[Cpt-Dingus]

I can confirm it exists in the boot entries after running the commands. What is happening here?

[jglathe]

Well then its "delete all boot entries" in uefi BIOS, select only the disk. Or, if you know which one works, leave that as the only one available. I have no real idea what the EFI boot tries, bot there are off things, depending on device.

[Cpt-Dingus]

Tried. Deleted everything except for wbm with efibootmgr, deleted ubuntu from /boot/efi/EFI, reran commands, no help

Can we maybe get in touch via matrix? A real time comm might be beneficial.

[jglathe]

go ahead

[jglathe]

Update to add: It works, but boy is the Acer UEFI firmware "special"... It appears to be able to only boot the first UEFI boot entry reliably. This must be grub , grub can boot Windows and Ubuntu.
The procedure described may work. Pay attentieon to the filesystem UUIDs, need to be different, grub OS-prober seems to be keen on interchanging them.

[lakatdev]

I installed this said image to an empty nvme disk by burning it there and later manually partitioning it for windows too (https://gist.github.com/caccialdo/3b0d0113489ecee456d94c1e9462d755) and copying windows files from the Windows 11 ARM ISO. Booted windows and installed the firmwares and drivers using windows update and just after that booted the Ubuntu image for the first time. It worked!

Now the main issue i have is the CPU fan not working at all and the laptop shutting down from throttling after continuous use. Not sure what to do with that yet.

[jglathe]

Actually, congratulations. 🚀 That fan issue will be sort of stay with us until we either can load the appropriate EC driver, or get more than rudimentary SoC support for x126* / x1p4* Purwa chips. This is actually an issue with this model, The Thinkbook 16 has very potent fans, also active on Linux. I know of this kind of behaviour from the Vivobook S15, apparently Acer, I thought I also heard it from the Yoga Slim 7. HP Omnibook has dynamic fans, too, but can also emergency switch off if the airflow isn't enough. Maybe there is a BIOS setting to change that behaviour... actually never looked.
What's the issue with the Purwa SoC: It is a scaled-down variant of Hamoa (x1e80100). They are drop-in pin compatible (apparently) , but the device trees are not the same. x1p42100.dtsi is derived from x1e80100.dtsi. Because there is a smaller GPU and fewer cores, it is also one tsens sensor bank less. The assignment (and interrupt routing of these), is probably different - that info is missing. The SoC has some internal firmware thermal management, so it will switch off before frying the chip. But maybe a delve into the firmware device trees can be useful for this assignment, come to think of it.
There's also good news: X1 has appeared as chip-id for Chromebooks. So we might benefit from their implementations, eventually.

[vpsha8888]

Hi,
I have installed this Acer Go specific image on a USB stick. I could boot my Acer SFG14-01 x1p42100 succesfully (type A as well as type C port). Wifi works. So I make this post from the same device. GPU shows software rendering. dmesg shows error about sound card:
[ 10.414005] snd-x1e80100 sound: ASoC: failed to instantiate card -2
[ 10.414498] snd-x1e80100 sound: probe with driver snd-x1e80100 failed with error -2

System Details Report

Report details

• Date generated: 2025-05-20 20:13:38

Hardware Information:

• Hardware Model: Acer Swift SFG14-01
• Memory: 16.0 GiB
• Processor:
• Graphics: Software Rendering
• Disk Capacity: 512.1 GB

Software Information:

• Firmware Version: V1.19
• OS Name: Ubuntu 24.10
• OS Build: (null)
• OS Type: 64-bit
• GNOME Version: 47
• Windowing System: Wayland
• Kernel Version: Linux 6.15.0-rc1-jg-3-qcom-x1e

Before finding this image, I tried boot the iso image with with dtb modified etc. But aborted after 4 hours in my longest attempt.
Thanks a lot for making this image, which works fine on every boot. But many thing not work including fan as already pointed out in previous post.
With respect to the currently used dtb / kernel what is involved in making the fan works at least? Thanks.

[jglathe]

Congrats for bringing it up! The error message is due to the missing topology file, not sure if it is there, (like, upstreamed to audioreach-topology) but It I should take a look if you're willing to test and risk your speakers.
Fan is quite an issue: Probably controlled by EC, without the hardware for probing this could be a challenge.
What I see from your data: You could switch to a newer kernel (on my Google Drive) and you can switch the GUI to using X11, which is faster with llvmpipe software rendering.
Regarding fans: Without additional data from the vendor, not much hope. It usually improves when a capable dev has such a device.

[JamiKettunen]

This may sound a bit random but could someone with this laptop check out https://github.com/TravMurav/dtbloader#adding-new-devices and run the script (with qcom/x1p42100-acer-swift-go14-01.dtb specified)? That would more or less finalize support there for released X1 devices with devicetrees in Ubuntu/LKML/mainline. Thanks :)

[lakatdev]

is this output what you need?

$ sudo scripts/describe_hw.sh -d "qcom/x1p42100-acer-swift-go14-01.dtb"
Loading…: 0%
Loading…: 1%
Loading…: 2%
Loading…: 5%
// SPDX-License-Identifier: BSD-3-Clause

#include <efi.h>
#include <device.h>

static EFI_GUID acer_swift_go_14_ai_hwids[] = {
{ 0x795d7de3, 0x0e1e, 0x5976, { 0xb0, 0xd2, 0x22, 0x16, 0x37, 0x5d, 0xf4, 0x42 } },
{ 0x327e7f1e, 0x4a10, 0x5fe1, { 0x8b, 0x4f, 0x3c, 0x23, 0x5e, 0xa1, 0x28, 0xd9 } },
{ 0x627edbb7, 0xfbdb, 0x5de7, { 0x83, 0x82, 0xf7, 0xab, 0xcb, 0xf9, 0x92, 0x8a } },
{ 0x79dd27da, 0x6d6b, 0x57da, { 0x8f, 0xba, 0xb9, 0x7b, 0x48, 0x9b, 0xe6, 0x1d } },
{ 0x3bd63b6b, 0xdfa3, 0x5314, { 0x98, 0x46, 0x76, 0x36, 0x3a, 0xe4, 0x94, 0x45 } },
{ 0x03f8cc3c, 0x5d61, 0x5ea0, { 0xae, 0xb1, 0xb3, 0x14, 0x25, 0x28, 0xbc, 0xe2 } },
{ 0x3f49141c, 0xd8fb, 0x5a6f, { 0x8b, 0x4a, 0x07, 0x4a, 0x23, 0x97, 0x87, 0x4d } },
{ 0x7c107a7f, 0x2d77, 0x51aa, { 0xae, 0xf8, 0x8d, 0x77, 0x7e, 0x26, 0xff, 0xbc } },
{ 0x2e71fd01, 0xd5f5, 0x5c67, { 0x9a, 0xbc, 0xc4, 0x7c, 0x10, 0xf0, 0x86, 0x26 } },
{ 0x6abb9980, 0x4909, 0x5d87, { 0x95, 0xa4, 0x87, 0x11, 0x29, 0x04, 0x73, 0x02 } },
{ 0xa07b1705, 0xc904, 0x5a73, { 0xb9, 0x7a, 0xf4, 0x6b, 0xc2, 0xdd, 0x84, 0x00 } },
{ 0x94782273, 0xeb4e, 0x57d4, { 0xb9, 0xc1, 0xbf, 0xc8, 0x85, 0x88, 0xe6, 0xc0 } },
{ 0x92dcc94d, 0x48f7, 0x5ee8, { 0xb9, 0xec, 0xa6, 0x39, 0x3f, 0xb7, 0xa4, 0x84 } },
{ 0x32f83b0f, 0x1fad, 0x5be2, { 0x88, 0xbe, 0x5a, 0xb0, 0x20, 0xe7, 0xa7, 0x0e } },
{ 0x1e301734, 0x5d49, 0x5df4, { 0x9e, 0xd2, 0xaa, 0x1c, 0x0a, 0x9d, 0xdd, 0xda } },
{ 0xe66990c4, 0xb390, 0x59c3, { 0xa5, 0x80, 0x14, 0x9c, 0x73, 0xcf, 0xe5, 0x85 } },
{ 0x69670c5e, 0x509d, 0x5193, { 0x9f, 0x0a, 0x88, 0x2a, 0x00, 0x18, 0x0b, 0x1a } },
{ 0x86654360, 0x65f0, 0x5935, { 0xbc, 0x87, 0x81, 0x10, 0x2c, 0x6a, 0x02, 0x2b } },
{ }
};

static struct device acer_swift_go_14_ai_dev = {
.name = L"Acer Swift Go 14 AI",
.dtb = L"qcom\x1p42100-acer-swift-go14-01.dtb",
.hwids = acer_swift_go_14_ai_hwids,
};
DEVICE_DESC(acer_swift_go_14_ai_dev);

[jglathe]

@lakatdev Wow that was fast, thanks. Can you also post the raw output of sudo fwupdtool hwids ? This would be needed for other methods like hwid2json.py (UKI) and stubble.

[lakatdev]

Loading…                 [*                                      ]
Computer Information
--------------------
BiosVendor: Insyde Corp.
BiosVersion: V1.03
BiosMajorRelease: 0
BiosMinorRelease: 0
FirmwareMajorRelease: 01
FirmwareMinorRelease: 03
Manufacturer: Acer
Family: Swift Go 14 AI
ProductName: Swift SFG14-01
ProductSku: 
EnclosureKind: a
BaseboardManufacturer: SX1
BaseboardProduct: Bluetang_SX1
Hardware IDs
------------
{795d7de3-0e1e-5976-b0d2-2216375df442}   <- Manufacturer + Family + ProductName + ProductSku + BiosVendor + BiosVersion + BiosMajorRelease + BiosMinorRelease
{327e7f1e-4a10-5fe1-8b4f-3c235ea128d9}   <- Manufacturer + Family + ProductName + BiosVendor + BiosVersion + BiosMajorRelease + BiosMinorRelease
{627edbb7-fbdb-5de7-8382-f7abcbf9928a}   <- Manufacturer + ProductName + BiosVendor + BiosVersion + BiosMajorRelease + BiosMinorRelease
{79dd27da-6d6b-57da-8fba-b97b489be61d}   <- Manufacturer + Family + ProductName + ProductSku + BaseboardManufacturer + BaseboardProduct
{3bd63b6b-dfa3-5314-9846-76363ae49445}   <- Manufacturer + Family + ProductName + ProductSku
{03f8cc3c-5d61-5ea0-aeb1-b3142528bce2}   <- Manufacturer + Family + ProductName
{3f49141c-d8fb-5a6f-8b4a-074a2397874d}   <- Manufacturer + ProductSku + BaseboardManufacturer + BaseboardProduct
{7c107a7f-2d77-51aa-aef8-8d777e26ffbc}   <- Manufacturer + ProductSku
{2e71fd01-d5f5-5c67-9abc-c47c10f08626}   <- Manufacturer + ProductName + BaseboardManufacturer + BaseboardProduct
{6abb9980-4909-5d87-95a4-871129047302}   <- Manufacturer + ProductName
{a07b1705-c904-5a73-b97a-f46bc2dd8400}   <- Manufacturer + Family + BaseboardManufacturer + BaseboardProduct
{94782273-eb4e-57d4-b9c1-bfc88588e6c0}   <- Manufacturer + Family
{92dcc94d-48f7-5ee8-b9ec-a6393fb7a484}   <- Manufacturer + EnclosureKind
{32f83b0f-1fad-5be2-88be-5ab020e7a70e}   <- Manufacturer + BaseboardManufacturer + BaseboardProduct
{1e301734-5d49-5df4-9ed2-aa1c0a9dddda}   <- Manufacturer
Extra Hardware IDs
------------------
{e66990c4-b390-59c3-a580-149c73cfe585}   <- Manufacturer + Family + ProductName + ProductSku + BiosVendor
{69670c5e-509d-5193-9f0a-882a00180b1a}   <- Manufacturer + Family + ProductName + BiosVendor
{86654360-65f0-5935-bc87-81102c6a022b}   <- Manufacturer + BiosVendor

have fun

[jglathe]

ubuntu/stubble#37 (comment)

[seqre]

Hello! I have Acer Swift Go 14 AI and I used your new ISO installer to try to install Linux on it. It successfully booted into the live environment using my machine's profile and installed system on the machine. However, when I try to boot into it, it starts loading, starts init system, and after printing few more log lines, it reboots.

Here's dmesg.log from live boot in case it's useful. It shows a bunch of different errors, but it booted nevertheless, so I'm not sure if they matter.

I recorded in slow-mo the screen of laptop booting up from the disk (I don't know of better way to save that 😅 ):

• First line I saw was [ 23.605646] kernel: tun: Universal TUN/TAP device driver, 1.6
• Until [ 23.606031] kernel: AppArmor: AppArmor sha256 policy hashing enabled line, everything more or less the same as live-boot one
• Then there was [Firmware Bug]: Unable to handle paging request in EFI runtime service and backtrace
• Then there were:

efi: EFI Runtime Services are disabled!
integrity: Couldn't get size: Device Error (0x0000000000000007)
integrity: MODSIGN Couldn't get UEFI db list
integrity: Couldn't get size: Device Error (0x0000000000000007)
integrity: Couldn't get UEFI dbx list

• Then it continued as if nothing happened from [ 23.606037] kernel: ima: secureboot mode unknown, no efi as before
• It went until the [ 23.608394] kernel: clk: Not disabling unused clocks message, after which the screen went blank and machine rebooted

@jglathe, is there anything I could try to make it work or do you need any extra info/help from me to troubleshot that? I'd love to get this machine finally working and I'm willing to help, but I need some pointers as I've never worked this low-level before.

[jglathe]

Hi, the issue is probably (not quite sure) that the installation relies on a kernel that has stubble, and that stubble knows the hwid and has a matching dtb - its all linked into the kernel. The Swift Go 14 is not in the stubble list afaik. To work around it you need to add a line to the grub menu entries for Linux:
devicetree /boot/x1p42100-acer-swift-go14-01.dtb
Also, you need to copy over the dtbs from the root of the KERNEL_PKG partition on the ISO to your /boot directory to have it accessible. These dtbs are "hacked" as in they have another compatible marker for the scm driver to pose as a Thinkpad T14s, enabling scm and efivars, required for installation. That works on kernels that don't know the newer models yet.

In general, I just finished the first 6.17 build with Ubuntu Concept 6.17 build, including stubble. Learned a few things, will add all "new" dtbs where I have dt and hwid to it. With that it will get easier, eventually. Also, this solves update issues which stem from hacking ub the generated grub.cfg.

[seqre]

Thank you for the prompt response and that was it! Once I modified the /boot/grub/grub.cfg with the devicetree entry the system booted up. However, as you mentioned, the next time grub.cfg is regenerated it's going to disappear. Would it make sense to modify /etc/grub.d/10_linux or create another script to take care of that? Also, could we modify Calamares to copy that .dtb file automatically for each profile?

When booted, I've noticed that wireless networking and USB ports stopped working. Looking at dmesg output, I see the following:

• a lot of Fixed dependency cycle(s) with messages
• again the Unable to handle paging request in EFI runtime service error and stack trace
• again the integrity errors
• bunch of initialization/device link creation errors
• continuously happening Direct firmware load for qcom/gen71500_sqe.fw failed with error -2

Here's dmesg.log from successful boot up from the disk. If there's anything I can do, please let me know!

[jglathe]

Hi @seqre , well most of this is a bit out of my reach.

• fixed dependency cycle spam: can't change that
• Unable to handle paging request in EFI runtime service: never seen that. But I always start with efi=noruntime as cmdline parameter.
• Integrity Errors: can't change that, always been there.
• pmic-glink errors: as reliable as the seasons. I had one or two kernels (in the 6.7 era) where this was gone, it is constantly there and I don't understand why it's never fixed.
• direct firmware load of gpu firmware: argh. This one is probably due to the ISO being 25.04, where Mesa doesn't have X1-45 support. Upgrading to 25.10 should fix this. But you need to have fetched all the right firmware from windows for this. Seen that its a Swift 14 Go, this is not yet supported by qcom-firmware-extract. I will push a version for it to https://github.com/jglathe/wdk2023_fw_fetch. Of course you could derive it from the Ideapad 5 branches yourself, too.

In general regarding boot config hassles: Working on the switch to kernels with stubble that detect the model and select the linked dtb (UKI style). Toolchain is working, todo is to integrate the dtb selection during build based on available hwid sets from stubble. This would be the most efficient way to solve it.