asus-expertbook-linux

Linux compatibility patches for the 2026 ASUS ExpertBook Ultra (B9406CAA) — packaged as a tracked, versioned, reversible patcher.

🌐 Documentation site · Quick install · Modules · Before / after · FAQ

---

Is this repo for me?

It's for you if all of the following are true. The single command below checks them in one go:

curl -fsSL https://raw.githubusercontent.com/burakgon/asus-expertbook-linux/main/scripts/check-hardware.sh | bash

Check	Expected	Why it matters
Laptop model (DMI)	ASUS EXPERTBOOK B9406CAA	All fixes are scoped to this exact subsystem ID
CPU family	Intel Core Ultra Series 3 (Panther Lake)	Required for the xe driver / iwlmld paths
Touchpad	PixArt I²C-HID 093A:4F05 (ACPI ASCP1D80)	The pressure-axis quirk applies here
Audio codec	Cirrus CS42L43 + 2× CS35L56 (subsystem 1043:15e4)	Per-OEM speaker firmware needed
Wi-Fi card	Intel Wi-Fi 7 BE211 (8086:e440)	iwlmld-mode tunables apply here
Distro	Arch / CachyOS / any Arch-derivative	The patcher uses pacman and reads /etc paths Arch-style

If you're on a sibling model (104315d4 / 104315f4) and willing to test, see Adding a new module / model. If you're on a different distro, the modules themselves still apply — only the pacman-based package-install steps are Arch-specific.

What this actually fixes — before / after

Hardware	Symptom out of the box	After installing	Module
PixArt I²C-HID haptic touchpad 093A:4F05 (ACPI ASCP1D80)	Touchpad doesn't move the cursor. Kernel log spams kernel bug: Touch jump detected and discarded. libinput rejects every event.	Cursor responds to light touches like any normal laptop. Zero "Touch jump" lines.	touchpad-fix
Cirrus CS42L43 codec + 2× CS35L56 speaker amps (PCI subsystem 1043:15e4)	Speakers are completely silent. dmesg: cs35l56: FIRMWARE_MISSING, Calibration disabled. F1 mute LED stuck on.	Speakers play at any volume. dmesg: Calibration applied, Tuning PID: 0x23134.	audio-fix
Intel Wi-Fi 7 BE211 Panther Lake CNVi (8086:e440)	Wi-Fi 7 (802.11be / EHT) is unstable. EHT RX collapses to MCS0/NSS1, MLO sessions tear down, missed beacons spam, occasional Microcode SW error freezes.	EHT disabled (disable_11be=Y) → rock-solid Wi-Fi 6 / HE fallback (6 GHz, 160 MHz, ~2.1 Gbit/s verified). Zero beacon spam, no freezes.	wifi-fix
Samsung Display Corp eDP panel + Intel xe driver (Xe3 Panther Lake iGPU)	Internal panel goes black. kwin_wayland: Pageflip timed out! This is a bug in the xe kernel driver. eDP-1 wedges, only reboot recovers.	Internal display stable indefinitely. PSR / Panel Replay disabled cleanly at boot.	display-fix
Intel Core Ultra X7/X9 Panther Lake hybrid (P + E + LP-E cores)	Idle power 4–5 W, fans audible at idle, P-cores never deep-sleep.	Idle ≈ 2–2.5 W. Workload parks on a single LP-E core. P-cores reach C10.	intel-perf-fix
USB UVC webcam (+ idle Panther Lake NPU)	No AI camera effects. Windows Studio Effects (background blur, smart framing) doesn't exist on Linux out of the box.	CPU background blur via OBS + obs-backgroundremoval, exposed as a virtual camera ("AI Camera"). (NPU offload is not available in the OBS plugin on Linux — see the module's reality-check note.)	webcam-ai-fix
ASUS BIOS SLKB ACPI method (BIOS B9406CAA.304)	KDE keyboard-backlight slider does nothing — but the Fn hotkeys still work (the backlight is not dead). SLKB clamps OS-initiated 0..3 writes to Local0 = Zero, so KDE / brightnessctl / sysfs writes silently no-op.	(optional) KDE slider works: asusd translates kernel writes into the OEM-tested 0x100..0x103 range.	keyboard-backlight-fix

Nothing this repo installs is a band-aid in the bad sense. Every module uses the exact same upstream-recognised mechanism (udev hwdb, libinput quirks, modprobe.d, systemd-tmpfiles, NetworkManager dispatcher, ALSA UCM codec dirs) that distros use to support every other laptop. We just haven't been added to the canonical lists yet — the upstream-patches/ folder is the path to that.

Quick install

git clone https://github.com/burakgon/asus-expertbook-linux.git
cd asus-expertbook-linux
./patch.sh install-all
sudo reboot

After reboot:

./patch.sh status

You should see all seven modules up to date and their runtime checks green.

Or pick à la carte

./patch.sh list                          # see what's available
./patch.sh install touchpad-fix audio-fix
./patch.sh diff display-fix              # preview before installing
./patch.sh uninstall wifi-fix            # back out anytime

Or run the interactive menu

./patch.sh

Auto-elevates with sudo, lets you install / uninstall / diff / status by typing single letters. Numbered table, color-coded state, cached.

=== asus_expertboot_linux patcher ===

  #   Module                  Version  Installed State          Description
  ----------------------------------------------------------------------------------
  1   audio-fix               2.1.0    2.1.0     up to date     Speakers + mics + clean panel
  2   display-fix             1.1.2    1.1.2     up to date     xe Panel Replay / PSR lockup
  3   intel-perf-fix          1.1.0    1.1.0     up to date     thermald + intel-lpmd
  4   keyboard-backlight-fix  1.1.0    1.1.0     up to date     (optional) KDE backlight slider
  5   touchpad-fix            1.1.1    1.1.1     up to date     PixArt 093A:4F05 pressure quirk
  6   webcam-ai-fix           1.1.0    1.1.0     up to date     OBS CPU background blur
  7   wifi-fix                2.0.0    2.0.0     up to date     BE211: disable broken EHT

Actions
  i <num>    install / update module (idempotent — re-runs post hooks)
  u <num>    uninstall module
  d <num>    diff source vs installed (omit num for all)
  s <num>    detailed status (omit num for all modules)
  I          install all modules
  up         update all currently-installed modules
  U          uninstall all modules
  r          refresh
  q          quit

>

Modules

1. touchpad-fix — light-touch cursor

The bug — pressure axis mis-parsed by hid-multitouch

The kernel's HID descriptor parser inflates ABS_MT_PRESSURE max to 2601 (literally the Y-axis max value, suggesting a parser typo) for this PixArt haptic touchpad. Real hardware values top out around 1000. libinput's pressure thresholds are calibrated against the kernel-reported max, so real touches register at 1–6% of the bogus "max" — well below the activation threshold. Result: every motion is rejected as a "kernel bug: Touch jump."

$ sudo dmesg | grep "Touch jump" | wc -l
1873                                    ← without the module
0                                       ← with the module

The fix — udev hwdb pressure clamp + libinput quirk

File	Path	What it does
61-pixart-4f05-pressure-fix.hwdb	/etc/udev/hwdb.d/	Clamps EVDEV_ABS_18 (ABS_PRESSURE) and EVDEV_ABS_3A (ABS_MT_PRESSURE) to a sane range so libinput's pressure heuristics see usable values.
99-asus-expertbook-pixart-4f05.quirks (installs as local-overrides.quirks)	/etc/libinput/	Tells libinput to ignore the pressure axes entirely via AttrEventCode=-ABS_MT_PRESSURE;-ABS_PRESSURE. Same shape as the shipped Asus UX302LA quirk.

After install, libinput quirks list /dev/input/event9 confirms the quirk is loaded.

2. audio-fix — speakers, headphones, mics (HiFi UCM)

The bug — firmware, a UCM gap, and topology noise

1. The Cirrus CS35L56 speaker amps need per-OEM tuning firmware. As of linux-firmware-cirrus >= 20260519 it ships upstream for 1043:15e4; on anything older the amps boot FIRMWARE_MISSING and the bundled blobs fill in.
2. The card reports a combined sidecar-amp codec — spk:cs35l56+cs42l43-spk (or two spk: tags on older kernels). Stock alsa-ucm-conf 1.2.15.x has no UCM dir for it and its SpeakerCodec regex drops the trailing -spk, so alsaucm fails (codecs/cs35l56+cs42l43/init.conf: -2). WirePlumber then uses stereo-fallback, which plays to the Jack PCM (device 0), not the Speaker PCM (device 2) — silent speakers, even though aplay -D plughw:0,2 works.
3. The generic SOF topology declares an unused SSP2-BT hardware-offload PCM with no firmware blob; WirePlumber's probe of it spams the kernel log (~40% of all kernel errors at boot).

$ sudo dmesg | grep cs35l56
cs35l56 sdw:0:2:01fa:3556:01:0: FIRMWARE_MISSING                    ← without
cs35l56 sdw:0:2:01fa:3556:01:1: FIRMWARE_MISSING                    ← without
─────────────────────────────────────────────────────────────────────────
cs35l56 sdw:0:2:01fa:3556:01:0: Calibration applied                 ← with
cs35l56 sdw:0:2:01fa:3556:01:0: Tuning PID: 0x23134, SID: 0x470200  ← with

The fix — HiFi UCM + cs35l56 firmware (replaces the old pro-audio pin)

The proper fix is the upstream HiFi UCM, not a profile hack — named ports, headphone-jack auto-switching, working volume + mic-mute LED. It's upstream as of alsa-ucm-conf 1.2.16, so on 1.2.16+ this module installs only the firmware + the SSP2-BT drop-in; the UCM rows below are dropped in only as a fallback on alsa-ucm-conf < 1.2.16 (and the NoExtract pin is removed automatically once the package crosses 1.2.16).

File	Path	What it does
cs35l56-…-l2u{0,1}.{bin,wmfw}	/lib/firmware/cirrus/	Per-OEM tuning + ROM 3.4.4→3.13.4 patch. Fallback — linux-firmware-cirrus >= 20260519 now ships these.
sof-soundwire.conf	/usr/share/alsa/ucm2/sof-soundwire/	Upstream alsa-ucm-conf master: fixes the SpeakerCodec regex to keep the -spk suffix. Pinned via NoExtract so a partial upgrade can't revert it (both only on alsa-ucm-conf < 1.2.16).
cs35l56+cs42l43-spk.conf, cs42l43-spk+cs35l56.conf	/usr/share/alsa/ucm2/sof-soundwire/	The Speaker device for the combined codec — routes playback to hw:,2 and the CS35L56 + CS42L43 amps.
cs42l43-spk+cs35l56-init.conf	/usr/share/alsa/ucm2/codecs/cs42l43-spk+cs35l56/	Combined codec init (control remap + LED attach). module.sh symlinks cs35l56+cs42l43-spk → this so both kernel names resolve.
52-disable-bt-sco-offload.conf	/etc/wireplumber/wireplumber.conf.d/	Disables the dead SSP2-BT offload PCM so its probe stops spamming the log. Bluetooth audio (A2DP/HFP) still works via the PipeWire software path.

The F1 speaker-mute LED can't be fixed from Linux — this laptop exposes no speaker-mute LED device, only platform::micmute (which the HiFi UCM drives).

3. wifi-fix — BE211: disable broken EHT, stabilize Wi-Fi 6

The bug — Wi-Fi 7 / EHT is broken on BE211

The core problem is 802.11be (EHT / Wi-Fi 7) itself on the Intel BE211 (8086:e440) under iwlwifi/iwlmld: the EHT RX path collapses to MCS0 / NSS1 and MLO sessions tear down, so the "Wi-Fi 7" link is slower and flakier than plain Wi-Fi 6. Not fixed upstream as of Linux 7.1-rc7. Two secondary irritants pile on: iwlmld defaults to power_scheme=2 (beacon-loss recovery churn), and the iwlwifi TX-segmentation offload bug can throw Microcode SW error under heavy traffic.

$ sudo dmesg | grep -E "missed beacons|Microcode SW error" | wc -l
2046                                    ← with EHT on
0                                       ← with disable_11be=Y (Wi-Fi 6 fallback)

The fix — disable broken EHT, keep a stable Wi-Fi 6 link

Core fix — drop the broken 802.11be layer so the radio runs as rock-solid Wi-Fi 6 (HE). Same approach Omarchy ships; verified at ~2.1 Gbit/s over 160 MHz 6 GHz HE here:

File	Path	What it does
iwlwifi-disable-eht.conf	/etc/modprobe.d/	options iwlwifi disable_11be=Y — disables EHT / Wi-Fi 7; the link falls back to stable Wi-Fi 6 / HE.

Secondary tunables (trim remaining HE-mode instability; they do not keep Wi-Fi 7 alive):

File	Path	What it does
iwlmld-active.conf	/etc/modprobe.d/	options iwlmld power_scheme=1 — disables driver-side power-save loop.
pcie-aspm-performance.conf	/etc/tmpfiles.d/	Write performance to /sys/module/pcie_aspm/parameters/policy at boot.
90-iwlwifi-no-offload	/etc/NetworkManager/dispatcher.d/	ethtool -K $iface tso off gso off gro off on every iwlwifi up event.

This is a deliberate Wi-Fi 7 → Wi-Fi 6 downgrade — EHT is the problem on this silicon. iwlwifi.bt_coex_active=Y is left alone, so Bluetooth keeps working.

4. display-fix — internal panel doesn't lock up

The bug — xe driver hangs Panel Replay handshake

The Samsung Display Corp panel in this laptop reports IEEE OUI 00:aa:01 in DPCD register 0x300 and supports Panel Replay Selective Update (Early Transport). The xe driver's PSR idle wait times out on this panel firmware:

xe 0000:00:02.0: [drm] Selective fetch area calculation failed in pipe A   # every boot
xe 0000:00:02.0: [drm] *ERROR* Timed out waiting PSR idle state
xe 0000:00:02.0: [drm] *ERROR* [CRTC:151:pipe A] DSB 0 timed out waiting for idle
kwin_wayland: Pageflip timed out! This is a bug in the xe kernel driver

Once the display engine wedges, only a reboot recovers it — modeset cycle, GPU GT0 reset, and runtime PSR-disable via debugfs all fail.

It's worse than a black panel. When the PSR2 selective-fetch path deadlocks the DSB during heavy compositing (a screen capture is enough to trigger it), it can take the whole kernel down — a silent hard hang with no oops, no MCE, and an empty pstore/BERT. That software-DSB hang is distinct from the Lunar Lake PMC-firmware crash, which does leave a BERT: [Hardware Error] record and is not cured by disabling PSR.

The fix — xe.enable_psr=0 on the kernel cmdline

File	Path	What it does
xe-disable-psr.conf	/etc/modprobe.d/	Belt-and-suspenders for late module load.
(managed block)	/etc/default/limine	Appends xe.enable_psr=0 xe.enable_psr2_sel_fetch=0 xe.enable_panel_replay=0 to the kernel cmdline. The post-install hook calls limine-update so the new params land in every kernel entry of /boot/limine.conf. Uninstall removes the block cleanly.

This is structurally identical to the per-device entry the upstream drm-intel-next branch is growing for Dell XPS 14/16. Our upstream-patches/0001 ports the same approach to a proper intel_dpcd_quirks[] entry — once merged, this module becomes a no-op and can be uninstalled.

There is no dedicated upstream tracker for this Panther Lake PSR2 selective-fetch / DSB hang; it's reproduced locally on linux-cachyos 7.0.11 and linux-cachyos-rc 7.1-rc7. (drm/xe #7513 — "Lunar lake, rare shutdown under load" — is a distinct Lunar Lake PMC-firmware bug, not this one.) No fix is merged on any current kernel, so the cmdline workaround is still required.

5. webcam-ai-fix — Linux equivalent of Windows Studio Effects

The gap — no Linux equivalent shipped on Panther Lake "AI PC" laptops

Windows Studio Effects on Copilot+ PCs runs background blur, smart framing, eye-contact correction, and voice focus on the NPU. None of these are shipped on Linux out of the box, even though the Intel Panther Lake NPU itself is fully supported by the kernel (intel_vpu driver, /dev/accel/accel0 exposed) and the userspace stack (OpenVINO 2026, level-zero) is available via the AUR.

Without this module the NPU sits idle, the webcam feed has no AI processing, and there's no virtual-cam target for video chat apps to read from.

The fix — OBS pipeline + virtual cam + ML segmentation plugin

File / package	Source	What it does
v4l2loopback.conf	/etc/modules-load.d/	Auto-load v4l2loopback at boot
v4l2loopback-options.conf	/etc/modprobe.d/	Persistent device config (devices=1 video_nr=10 card_label='AI Camera' exclusive_caps=1)
v4l2loopback-dkms package	extra	Kernel module providing the virtual cam
obs-studio package	extra	Capture + filter graph + virtual-cam writer
obs-backgroundremoval package	AUR	ML segmentation OBS plugin (ONNX models). CPU-only on Linux — its execution providers are CUDA / ROCm / MIGraphX; there is no OpenVINO/NPU path in the OBS plugin.

The user is also added to the render group as defensive future-proofing for stricter NPU device permissions. /dev/accel/accel0 ships world-writable today.

After install, the user opens OBS, adds a Video Capture Device source pointing at the real webcam, attaches the Background Removal filter, and starts the virtual camera. Any video chat app then sees the processed feed as "AI Camera".

Reality check — no NPU offload in OBS on Linux. obs-backgroundremoval has no OpenVINO/NPU execution provider on Linux (installing openvino does not add an "NPU" device); the filter runs on the CPU — fine for 720p30 background blur. For an actual NPU route see ericjchang/linux-studio-effects (OpenVINO + v4l2loopback), but it's validated on Arrow Lake, not yet Panther Lake, and installs via git + pip. Also: if another tool already uses v4l2loopback (e.g. linuxdrop on video_nr=20), this module's global options line collides — share one devices=2 video_nr=10,20 config instead.

6. intel-perf-fix — Panther Lake idle / thermal

The bug — kernel-default thermal throttle and idle scheduling are coarse on Panther Lake

Without a userspace thermal daemon, the kernel governor's only lever is "cap CPU frequency". On Panther Lake's hybrid topology (P-cores + E-cores + LP-E cores), a P/E-aware throttle is far smarter — it can park work on slower cores instead of slowing everything down.

Without intel-lpmd, idle work spreads across multiple cores; with it, all idle work concentrates on a single LP-E core and the P-cores deep-sleep.

The fix — install + enable thermald and intel-lpmd

Package	Source	Service	Effect
thermald	extra repo	thermald.service	P/E-core-aware thermal throttle.
intel-lpmd	extra / cachyos repo	intel_lpmd.service	Parks idle work on LP-E core, lets P-cores deep-sleep.

Both coexist with the existing power-profiles-daemon (PPD handles user profile, thermald handles thermal, intel-lpmd handles idle topology).

This module ships no payload files — it's purely package install + service enable in the post-install hook. The patcher tracks it the same way it tracks file-based modules (versioned, idempotent, status-checked).

7. keyboard-backlight-fix — (optional) restore software/KDE backlight control

The backlight is not dead — the Fn hotkeys toggle it fine (handled by the EC in hardware, bypassing the buggy ACPI path). This module is optional: it only restores software control (the KDE slider / brightnessctl / sysfs).

The bug (software side) — ASUS BIOS clamps OS-initiated brightness writes to zero

The B9406CAA BIOS (B9406CAA.304) ships a broken SLKB ACPI method. Disassembled from the live DSDT:

Method (SLKB, 1, NotSerialized) {
    If    ((Arg0 >= 0x0100) && (Arg0 <= 0x0106)) { Local0 = (Arg0 - 0x0100) }
    ElseIf((Arg0 >= 0x80)   && (Arg0 <= 0x83))   { Local0 = (Arg0 - 0x80) * 0x21 ... }
    ElseIf((Arg0 >= Zero)   && (Arg0 <= 0x03))   { Local0 = Zero }   // ← BUG
    STBC (Zero, Local0)
    Return (One)
}

The mainline asus-wmi Linux driver writes the standard 0..3 kernel range, which hits the third branch — and it unconditionally clamps Local0 to zero before passing to STBC (the EC command emitter). End result: every KDE / brightnessctl / direct /sys write is silently turned into "set brightness 0", and the keyboard backlight stays off.

The OEM-tested 0x100..0x103 range works correctly. Verified by hand via acpi_call: invoking \_SB.PC00.LPCB.EC0.SLKB 0x103 lights the backlight.

The fix — let the asusd userspace daemon translate the value range

File / package	Path	What it does
xyz.ljones.Asusd.service	/usr/share/dbus-1/system-services/	D-Bus activation entry for asusd. The asusd.service systemd unit is Type=dbus, but ASUS doesn't ship the matching D-Bus service file — without it nothing ever auto-starts the daemon.
acpi_call.conf	/etc/modules-load.d/	Auto-load the acpi_call kernel module so /proc/acpi/call is available for direct ACPI invocations during debugging.
asusctl package	extra repo	Provides the asusd daemon.
acpi_call-dkms package	AUR	Optional. Kernel module exposing /proc/acpi/call.
/etc/asusd/ directory	(created in post_install)	asusd refuses to start without it; the package leaves it absent.

asusd translates standard kernel-level brightness writes into the OEM-tested 0x100..0x103 range before invoking ACPI, side-stepping the buggy branch. KDE PowerDevil's keyboard-brightness control then reaches the EC correctly.

How it works

The whole project is a small bash module manager (patch.sh, ~500 lines) plus folders. Each subfolder containing a module.sh is a discoverable module:

asus-expertbook-linux/
├── patch.sh                    # the manager
├── audio-fix/
│   ├── module.sh               # manifest: files + hooks + status check
│   ├── README.md
│   └── …                       # payload files
├── display-fix/  …
├── intel-perf-fix/  …
├── keyboard-backlight-fix/  …
├── touchpad-fix/  …
├── webcam-ai-fix/  …
├── wifi-fix/  …
├── upstream-patches/           # submission-ready upstream patches
│   └── 0001…0003.patch
├── docs/                       # the GitHub Pages site
└── scripts/
    └── check-hardware.sh       # one-shot compatibility check

A module's manifest declares files (source → destination), an optional post-install hook, an optional status-check function, and a version. The patcher records the installed version under /var/lib/asus_expertboot_patcher/<module>.version so subsequent operations know whether each module is up to date, update available, partial, untracked, or not installed.

Command	Effect
./patch.sh	Interactive menu; auto-elevates to root via sudo.
./patch.sh list	Quick table of every module + its current state.
./patch.sh status [module…]	Detailed status: file presence + runtime probe + service state.
./patch.sh install [module…]	Idempotent install. Re-running applies any source updates.
./patch.sh update [module…]	Alias for install.
./patch.sh uninstall [module…]	Remove files + run uninstall hook.
./patch.sh diff [module…]	Show what would change before installing.
./patch.sh install-all	Install every discoverable module.
./patch.sh update-all	Re-install only modules that aren't up to date.
./patch.sh uninstall-all	Tear down everything cleanly.

Kernel & distro compatibility

• Linux 6.18+ for the haptic-touchpad kernel parser, the new iwlmld Wi-Fi 7 op-mode, the xe driver Panther Lake bringup, and the cs35l56 driver. Anything older won't even probe most of this hardware.
• Tested on: linux-cachyos 7.0.x, linux-cachyos-rc through 7.1-rc7 (the display-fix PSR hang persists on all of them). Should work on linux-lts 6.18.x and linux 7.0.x Arch builds.
• Distros: Arch and Arch derivatives (CachyOS, EndeavourOS, Manjaro) all use the same /etc/udev/hwdb.d, /etc/libinput, /etc/modprobe.d, /etc/wireplumber/wireplumber.conf.d paths the modules write to.
• Bootloader assumption (display-fix): limine via limine-mkinitcpio-hook, where /etc/default/limine is the source of truth. If you use systemd-boot or GRUB, the module's cmdline-injection hook needs swapping; the modprobe.d half still works.

Adding a new module or model

Drop a folder containing a module.sh next to patch.sh. The patcher discovers it automatically. The smallest example is touchpad-fix/module.sh:

MODULE_NAME="my-fix"
MODULE_DESC="One-line description"
MODULE_VERSION="1.0.0"

MODULE_FILES=(
  "src-relative-to-module-dir:/absolute/dst/path"
)

module_post_install()   { …; }   # optional
module_post_uninstall() { …; }   # optional
module_status_extra()   { …; }   # optional

Sibling-model contributions for 1043:15d4 and 1043:15f4 ExpertBook Ultra variants are very welcome — open a PR with your subsystem ID's firmware blobs (if cs35l56 is the same chip family) and any DMI tweaks needed.

Upstream submissions

The upstream-patches/ folder ships three patches that turn each module into a permanent upstream entry:

#	Tree	Replaces
0001	drivers/gpu/drm/i915/display/intel_quirks.c	display-fix's cmdline workaround
0002	sound/soc/intel/boards/sof_sdw.c	most of audio-fix (combined-codec UCM routing)
0003	libinput/quirks/30-vendor-pixart.quirks	touchpad-fix's libinput override

All three dry-run apply cleanly against current torvalds/linux master / drm-intel-next / libinput main. See upstream-patches/README.md for hardware identifiers, mailing list addresses, and submission instructions.

License

MIT for the code (scripts, configs, patches).

Firmware files redistributed under audio-fix/ come verbatim from upstream linux-firmware under their original Cirrus Logic redistribution license. See NOTICE.

FAQ

Does this work on similar 2026 ExpertBook Ultra models?

Most of it transfers. The audio-fix firmware blobs are matched on PCI subsystem 1043:15e4 (this exact laptop). Sibling subsystems 104315d4 and 104315f4 ship different per-OEM tuning files in upstream linux-firmware. The touchpad-fix, wifi-fix, display-fix, intel-perf-fix, webcam-ai-fix, and keyboard-backlight-fix modules are hardware-agnostic or match by family-level identifiers and apply more broadly.

PRs adding module.sh entries for sibling models are welcome.

Does this break Bluetooth?

No. The Wi-Fi module deliberately leaves iwlwifi.bt_coex_active=Y alone. Bluetooth audio, HID, and file transfer keep working as usual.

Does this downgrade Wi-Fi 7?

Yes — on purpose. 802.11be / EHT is broken on the BE211 (RX collapses to MCS0/NSS1, MLO tears down), so wifi-fix disables it (disable_11be=Y) and the link runs as stable Wi-Fi 6 / HE instead — ~2.1 Gbit/s over 160 MHz 6 GHz here, faster in practice than the flaky EHT link. Drop the module (or set disable_11be=N) once Intel fixes the iwlwifi EHT path upstream.

What about the F1 mute LED?

The HiFi UCM (active since audio-fix v2.0.0, and upstream in alsa-ucm-conf 1.2.16) drives the mic-mute LED (platform::micmute) correctly. The speaker-mute LED (F1) stays in its EC default state because this laptop exposes no speaker-mute LED device to Linux at all — there's nothing for the UCM SetLED hook to bind to. It's a missing-device limitation, not a profile issue.

Why not just upstream all of this and skip the repo?

That's the goal — see upstream-patches/. Two of the audio pieces already landed: alsa-ucm-conf 1.2.16 ships the cs42l43-spk+cs35l56 codec dir and linux-firmware-cirrus >= 20260519 ships the OEM blobs, so on an up-to-date system audio-fix is already down to a single topology-noise drop-in. As the kernel/quirk patches in upstream-patches/ land and your distro picks them up, every module here becomes deletable.

How do I test changes before installing?

./patch.sh diff <module>          # show before/after on every file the module manages

Output marks each file as unchanged / would update / would create with a coloured unified-diff for the changed ones.

Acknowledgements

• linux-firmware for the upstream CS35L56 OEM tuning blobs.
• alsa-ucm-conf for the shipped cs35l56, cs42l43, and cs42l43-dmic codec dirs that the combined cs42l43-spk+cs35l56/init.conf borrows from.
• Omarchy for surfacing how Panther Lake bring-up looks on the Hyprland side and which userspace daemons (thermald + intel-lpmd) are worth installing.
• The libinput project for the Asus UX302LA quirk pattern that touchpad-fix mirrors.