https://bugzilla.kernel.org/show_bug.cgi?id=220295
Bug ID: 220295
Summary: [REGRESSION] Internal keyboard on Machenike L16P (AMD
Ryzen 7 7735H) fails on recent kernels. Workaround
"acpi=off" broken in newer kernels.
Product: ACPI
Version: 2.5
Hardware: AMD
OS: Linux
Status: NEW
Severity: high
Priority: P3
Component: Other
Assignee: [email protected]
Reporter: [email protected]
Regression: No
Hello Kernel Development Team,
I am reporting a severe hardware compatibility issue and a subsequent kernel
regression affecting the internal keyboard of my Machenike L16P laptop. After
extensive troubleshooting across multiple distributions and kernel versions, it
has become clear that this requires kernel-level intervention.
**I. System Information**
=========================
* **Laptop Model:** Machenike L16P
* **CPU:** AMD Ryzen 7 7735H with Radeon Graphics
* **BIOS:** American Megatrends (AMI) N.1.14HAI04, dated 04/28/2023
* **Internal Keyboard:** Identified as "AT Translated Set 2 keyboard" on
i8042/serio0.
**II. Core Problem & Symptom**
============================
The internal keyboard is non-functional upon booting any recent Linux
distribution. The boot process completes, and the system is otherwise usable
with an external USB keyboard.
The primary symptom observed in the `dmesg` logs during a failed boot is the
i8042 controller's inability to initialize the keyboard port:
[ 2.900808] atkbd serio0: Failed to deactivate keyboard on isa0060/serio0
[ 3.307091] atkbd serio0: Failed to enable keyboard on isa0060/serio0
[ 3.307190] input: AT Translated Set 2 keyboard as
/devices/platform/i8042/serio0/input/input7
This sequence indicates the kernel detects the keyboard but fails during the
activation handshake.
**III. The Crucial Finding: A Kernel Regression**
=============================================
The key discovery is a behavioral change between kernel versions, which points
to a clear regression. This regression has removed the only known workaround.
* **Behavior on Older Kernel (Parrot OS, based on Debian, Kernel ~6.1):**
* **Normal Boot:** Keyboard does NOT work (as described above).
* **Boot with `acpi=off` parameter:** The internal keyboard **WORKS
PERFECTLY.** This strongly suggested a faulty ACPI implementation in the BIOS
was the root cause.
* **Behavior on Newer Kernel (Manjaro Linux Live USB, Kernel ~6.9):**
* **Normal Boot:** Keyboard does NOT work.
* **Boot with `acpi=off` parameter:** The internal keyboard **STILL DOES
NOT WORK.** The system boots, but the keyboard remains unresponsive.
This demonstrates that a change introduced between these kernel versions has
broken the `acpi=off` workaround, leaving this hardware with no method to
enable the internal keyboard on up-to-date systems.
**IV. Comprehensive Troubleshooting Steps (All Failed on Newer Kernels)**
=====================================================================
To isolate the issue, a wide range of diagnostic steps were taken. All of the
following failed to resolve the issue on the newer Manjaro kernel:
* **BIOS/UEFI Configuration:**
* The BIOS is extremely limited, with no access to IOMMU, SVM, Fast Boot,
or advanced ACPI settings.
* Secure Boot was disabled.
* The `AMD fTPM switch` was disabled. This had no effect on the
keyboard's status.
* **Kernel Boot Parameters (i8042/serio):**
* An exhaustive list of `i8042` parameters were tested individually and
in combination, including:
* `i8042.reset`
* `i8042.nopnp`
* `i8042.nomux`
* `i8042.dumbkbd`
* `i8042.probe_defer`
* `i8042.direct`
* **Kernel Boot Parameters (Module Blacklisting):**
* `dmesg` revealed that the `asus_wmi` module was being loaded
unnecessarily, likely due to the faulty ACPI data.
```
[ 7.952190] asus_wmi: ASUS WMI generic driver loaded
[ 8.009515] asus_wmi: ASUS Management GUID not found
```
* Booting with `module_blacklist=asus_wmi` did not solve the keyboard
issue.
* **ACPI OSI Strings:**
* Various ACPI OSI strings (e.g., `acpi_osi='Windows 2020'`) were tested
without success. A revealing message in the log confirms a general ACPI
firmware bug:
```
[ 0.311402] ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored
```
**V. Conclusion & Request**
===========================
The evidence strongly points to a combination of a faulty BIOS ACPI
implementation on the Machenike L16P and a kernel regression that has nullified
the only effective workaround (`acpi=off`).
The ideal solution would be a kernel patch that correctly handles the ACPI
quirks of this specific hardware, or at least restores the functionality of the
`acpi=off` workaround for the i8042 driver.
I am prepared to provide full `dmesg` logs from both "working" (older kernel
with `acpi=off`) and "non-working" (newer kernel) states, as well as a full
`acpidump`, if required.
Thank you for your dedication and for looking into this complex issue.
--
You may reply to this email to add a comment.
You are receiving this mail because:
You are watching the assignee of the bug.
_______________________________________________
acpi-bugzilla mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/acpi-bugzilla
