Support for BIOS V2.51 (2025): ACPI path changes (TBMD moved to VGBS) and Makefile improvements for Debian??? · Issue #1 · fstanis/chuwi-minibook · GitHub
You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Support for BIOS V2.51 (2025): ACPI path changes (TBMD moved to VGBS) and Makefile improvements for Debian??? #1
I've been testing your tools on a Chuwi MiniBook X N150 running Debian 13 with the latest BIOS V2.51 (04/16/2025). I found some breaking changes in this BIOS version compared to previous ones:
Tablet Mode Switch (ACPI Changes)
The traditional TBMD method used in minibook_ec is missing in this DSDT. Instead, it seems Chuwi has moved to a more modern Intel-style implementation:
Replacing TBMD with VGBS in minibook_ec_core.c is required for the tablet mode switch to be registered on this BIOS version.
2. Makefile Issues on Debian
The Makefiles for vbt_patch and goodix_ts currently fail on Debian because they try to fetch headers from kernel.org using the exact string from uname -r.
On Debian, the version string often includes a suffix (e.g., 6.12.86+deb13), which results in a 404 error when curl attempts to download intel_vbt_defs.h or goodix.c.
Suggestion: Add a fallback or a way to strip the distro-specific suffix from the version string in the Makefile.
3. VBT Patcher
In this environment, reading from /sys/kernel/debug/dri/0/i915_vbt failed with "VBT signature not found". I had to extract the VBT directly from ACPI using acpidump -n "VBT " -b to successfully apply the 60Hz patch.
Device Info:
Hey, thank you for the feedback @victorewik! I actually have the exact same BIOS version (see devices.md) so something else must be different.
Can you please share the full output of ./tools/detect-hardware.sh and sudo acpidump (with no parameters) so we can compare what's different between our two devices? It would be great if we can document the different variations CHUWI made.
Thank you for reporting the Debian issue too - I fixed it in #2, let me know if it builds with this.
Thank you for sharing! Our ACPI tables are practically identical, other than checksums, i.e. no difference in implemented methods. Hardware dump is identical as well.
One difference I noticed is that my card is drm:card1-DSI-1 while yours is drm:card0-DSI-1, but this is probably not relevant. Re-reading your original post, I noticed two things:
You said reading from /sys/kernel/debug/dri/0/i915_vbt failed - is this something you did manually? The script in tools actually reads it from a different path:
You mentioned changes to minibook_ec for the tablet mode switch - that module is just for thermal zones keyboard backlight etc. For tablet mode / screen rotation, you actually want the patched iio-sensor-proxy (link to how to install). If you use DE with Wayland, either it'll just work out of the box when you install that (KDE Plasma, GNOME) or you need to install iio-hyprland for most others.
Debian fix: I will try tomorrow but with kernel 6.12 mayby i change to cachy (i love so much debian so idk)
VBT Path: You are right. I was trying to read from /dri/0/ manually instead of using the full PCI path. That's why it failed for me. And i modify to use 60hz.
Tablet Mode: I didn't realize your minibook_ec doesn't expose a tablet_mode switch and that the logic is handled entirely by iio-sensor-proxy calculating the hinge angle. That makes much more sense now.
Orientation: Even though we have the same BIOS, my device reported "right-up" as the default orientation in landscape mode. I had to apply a custom ACCEL_MOUNT_MATRIX via HWDB/udev to get the auto-rotation to align correctly with GNOME's "Horizontal" setting.
@fstanis I tryed now in more specific for debian an i have some big problems
I've successfully finished a full installation of your tools on Debian 13 (Trixie/Stable). While I got it working, I encountered several issues that might be useful for improving compatibility with non-Arch distributions and newer hardware batches.
Build & DKMS Issues
Standard Debian uses GCC. The LLVM=1 flag in dkms.conf for the modules causes build failures. I had to manually strip them using:
find modules/ -name "dkms.conf" -exec sed -i 's/LLVM=1 //g' {} +
2. Service Installation Paths
On Debian, thermald and iio-sensor-proxy installation scripts default to paths that the system doesn't prioritize or where unit files are missing. I had to use specific flags to make them visible to systemd:
This is the most critical UX issue. I use video=DSI-1:panel_orientation=right_side_up in GRUB to ensure the boot logo and TTY are correctly oriented.
However, your patched iio-sensor-proxy seems to force a right-up orientation internally when in laptop mode. The result is a double rotation (90° + 90° = 180°).
Problem: When returning from Tablet to Laptop mode, the screen stays upside down.
Behavior: In Laptop mode, the screen still tries to rotate to "Left" and "Right" but never hits the correct "Horizontal".
Gnome: The rotation lock button in Gnome doesn't appear or doesn't work as expected because the proxy is forcing a state. monitor-sensor reports right-up when the device is flat on the table, while it should ideally be normal to align with the GRUB-rotated frame-buffer.
Backlight Sync Issue
The keyboard backlight slider in Gnome works, but it is not synchronized with the physical keyboard hotkey.
If I change the brightness via the physical key, the Gnome UI slider doesn't update.
The UI only "wakes up" if I move the Gnome slider first.
Single vs Dual Accelerometer
Despite having the same BIOS/Hardware, monitor-sensor only detects one accelerometer:
code Code
+++ iio-sensor-proxy appeared
=== Has accelerometer (orientation: right-up, tilt: vertical)
=== No ambient light sensor
Since the second accelerometer (base) isn't detected, the "Laptop mode" rotation lock isn't triggering, leaving the screen rotating even when used as a standard laptop.
System Info:
You shouldn't use video=DSI-1:panel_orientation=right_side_up if you use iio-sensor-proxy to rotate your screen - as you noticed, this will result in double rotation. GRUB (to my knowledge) doesn't support rotation the way Limine does and panel_orientation applies rotation to the framebuffer as a whole which isn't ideal if you want to combine it with rotation.
iio-sensor-proxy should be able to turn on the second accelerometer, but it won't be exposed via monitor-sensor, instead it'll expose orientation and tilt sensors and will also separately disable the keyboard in tablet mode. Is this working for you while monitor-sensor is running?
Regarding Debian GCC and paths, do you mind sending a PR? I can guess what the correct options are, but it's probably better if you could verify and send it.
I've prepared the changes to make the tools fully compatible with Debian 13 and other GCC-based distributions. Here is a summary of the improvements:
GCC Compatibility: Removed the hardcoded LLVM=1 from dkms.conf files. Most standard distributions use GCC for kernel modules. This change allows DKMS to use the system's default compiler.
Makefile Robustness: Modified the Makefiles for goodix_ts and vbt_patch to use a generic kernel URL. Debian's uname -r includes a +deb13 suffix which causes 404 errors on the kernel.org servers. The new logic ensures headers are downloaded correctly regardless of the distro suffix.
Standard FHS Paths: Updated the Thermal Daemon and IIO-Sensor-Proxy build instructions to use --prefix=/usr and /lib/systemd/system. This aligns with the Filesystem Hierarchy Standard used by Debian/Ubuntu for system services.
Keyboard Sync Logic: I'm also including the logic for the keyboard backlight synchronization that handles the BIOS timeout conflict we discussed.
Regarding keyboard sync - I tried to do it slightly differently in #8 by adding to the module itself. This should work since GNOME will autodetect the new /sys path.
You need to enable the feature using echo 'options minibook_ec poll_hw_changed=1' | sudo tee /etc/modprobe.d/minibook_ec.conf, then reboot (or remove / load the module) and try /sys/class/leds/minibook::kbd_backlight/brightness_hw_changed to read the current value (that's pushed via polling). Let me know if that works.
Activity
fstanis commentedon May 12, 2026
Hey, thank you for the feedback @victorewik! I actually have the exact same BIOS version (see devices.md) so something else must be different.
Can you please share the full output of
./tools/detect-hardware.shandsudo acpidump(with no parameters) so we can compare what's different between our two devices? It would be great if we can document the different variations CHUWI made.Thank you for reporting the Debian issue too - I fixed it in #2, let me know if it builds with this.
victorewik commentedon May 12, 2026
hardware_victorewik.txt
ID DRIVER NAME
dmi:vendor CHUWI Innovation And Technology(ShenZhen)co.,Ltd
dmi:product MiniBook X
dmi:bios DNN20A V2.51 04/16/2025
cpu:0 Intel(R) N150
pci:8086:46d4 i915 Intel Corporation Alder Lake-N [Intel Graphics]
pci:8086:461d proc_thermal_pci Intel Corporation Alder Lake Innovation Platform Framework Processor Participant
pci:8086:54ed xhci_hcd Intel Corporation Alder Lake-N PCH USB 3.2 xHCI Host Controller
pci:8086:54f0 iwlwifi Intel Corporation CNVi: Wi-Fi
pci:8086:54c4 sdhci-pci Intel Corporation Alder Lake-N eMMC Controller
pci:8086:54c8 snd_hda_intel Intel Corporation Alder Lake-N PCH High Definition Audio Controller
pci:1e4b:1202 nvme MAXIO Technology (Hangzhou) Ltd. NVMe SSD Controller MAP1202 (DRAM-less) (rev 01)
usb:8087:0026 usb
usb:1b0a:2bc9 usb Hy-Usb2.0-1*MIC
i2c: mxc4005
i2c:GDIX1002 Goodix-TS Goodix Capacitive TouchScreen
i2c:MDA6655 mxc4005
i2c:XXXX0000 i2c_hid_acpi XXXX0000:05 0911:5288 Touchpad
acpi:ACPI0003
acpi:ACPI000C
acpi:ACPI000E
acpi:GDIX1002
acpi:INT33A1
acpi:INT33D3
acpi:INT340E
acpi:INTC1041
acpi:INTC1046 Thermistor Ambient
acpi:INTC1057
acpi:INTC1070
acpi:INTC1099
acpi:MDA6655
acpi:MSFT0001
acpi:MSFT0101 tpm_crb TPM 2.0 Device
acpi:PNP0103
acpi:PNP0C02
acpi:PNP0C02
acpi:PNP0C02
acpi:PNP0C09 ec
acpi:PNP0C0A battery
acpi:PNP0C0C button
acpi:PNP0C0D button
acpi:PNP0C0E button
acpi:PRP00001
acpi:PRP00001
acpi:XXXX0000
blk:nvme0n1 driver AirDisk 512GB SSD (476 GB)
drm:card0-DSI-1 Internal panel 1200x1920
backlight:intel_backlight raw
hwmon:hwmon0 acpitz
hwmon:hwmon1 BAT0
hwmon:hwmon2 nvme
hwmon:hwmon3 ADP1
hwmon:hwmon4 iwlwifi_1
hwmon:hwmon5 coretemp
tpm:tpm0 v2
victorewik@debianewik:~/chuwi-minibook$
acpidump_victorewik.txt
In the end I went back to the minibook support I was using because I couldn't find the path to that sensor and the screen was inverted.
fstanis commentedon May 12, 2026
Thank you for sharing! Our ACPI tables are practically identical, other than checksums, i.e. no difference in implemented methods. Hardware dump is identical as well.
One difference I noticed is that my card is
drm:card1-DSI-1while yours isdrm:card0-DSI-1, but this is probably not relevant. Re-reading your original post, I noticed two things:You said reading from
/sys/kernel/debug/dri/0/i915_vbtfailed - is this something you did manually? The script in tools actually reads it from a different path:chuwi-minibook/tools/update-vbt-clock.sh
Line 9 in 563a0af
You mentioned changes to
minibook_ecfor the tablet mode switch - that module is just for thermal zones keyboard backlight etc. For tablet mode / screen rotation, you actually want the patchediio-sensor-proxy(link to how to install). If you use DE with Wayland, either it'll just work out of the box when you install that (KDE Plasma, GNOME) or you need to installiio-hyprlandfor most others.victorewik commentedon May 13, 2026
Hi @fstanis,
Thanks for the explanation!
Thanks for the amazing work!
victorewik commentedon May 15, 2026
@fstanis I tryed now in more specific for debian an i have some big problems
I've successfully finished a full installation of your tools on Debian 13 (Trixie/Stable). While I got it working, I encountered several issues that might be useful for improving compatibility with non-Arch distributions and newer hardware batches.
Standard Debian uses GCC. The LLVM=1 flag in dkms.conf for the modules causes build failures. I had to manually strip them using:
find modules/ -name "dkms.conf" -exec sed -i 's/LLVM=1 //g' {} +
2. Service Installation Paths
On Debian, thermald and iio-sensor-proxy installation scripts default to paths that the system doesn't prioritize or where unit files are missing. I had to use specific flags to make them visible to systemd:
This is the most critical UX issue. I use video=DSI-1:panel_orientation=right_side_up in GRUB to ensure the boot logo and TTY are correctly oriented.
However, your patched iio-sensor-proxy seems to force a right-up orientation internally when in laptop mode. The result is a double rotation (90° + 90° = 180°).
The keyboard backlight slider in Gnome works, but it is not synchronized with the physical keyboard hotkey.
Despite having the same BIOS/Hardware, monitor-sensor only detects one accelerometer:
code Code
+++ iio-sensor-proxy appeared
=== Has accelerometer (orientation: right-up, tilt: vertical)
=== No ambient light sensor
Since the second accelerometer (base) isn't detected, the "Laptop mode" rotation lock isn't triggering, leaving the screen rotating even when used as a standard laptop.
System Info:
I hope this feedback helps in making the project more robust for different setups!
fstanis commentedon May 19, 2026
You shouldn't use
video=DSI-1:panel_orientation=right_side_upif you useiio-sensor-proxyto rotate your screen - as you noticed, this will result in double rotation. GRUB (to my knowledge) doesn't support rotation the way Limine does andpanel_orientationapplies rotation to the framebuffer as a whole which isn't ideal if you want to combine it with rotation.iio-sensor-proxyshould be able to turn on the second accelerometer, but it won't be exposed viamonitor-sensor, instead it'll expose orientation and tilt sensors and will also separately disable the keyboard in tablet mode. Is this working for you whilemonitor-sensoris running?Regarding Debian GCC and paths, do you mind sending a PR? I can guess what the correct options are, but it's probably better if you could verify and send it.
victorewik commentedon May 24, 2026
Hi @fstanis,
I've prepared the changes to make the tools fully compatible with Debian 13 and other GCC-based distributions. Here is a summary of the improvements:
You will need free debian repos
and after install
I have verified these paths and options on a fresh Debian 13 installation and everything works perfectly.
https://github.com/victorewik/chuwi-minibook-Debian-
fstanis commentedon May 27, 2026
Regarding keyboard sync - I tried to do it slightly differently in #8 by adding to the module itself. This should work since GNOME will autodetect the new
/syspath.You need to enable the feature using
echo 'options minibook_ec poll_hw_changed=1' | sudo tee /etc/modprobe.d/minibook_ec.conf, then reboot (or remove / load the module) and try/sys/class/leds/minibook::kbd_backlight/brightness_hw_changedto read the current value (that's pushed via polling). Let me know if that works.fstanis commentedon Jun 8, 2026
I think all these issues are no resolved and fixes are merged, feel free to reopen if you find any other issues.