Package: linux-image-6.12.100+deb13-amd64
Version: 6.12.100-1
Followup-For: Bug #1143268
Hi Salvatore,
Bisection completed. The first bad commit is:
|dcc1bce68aec19333ed84d7f315a67e6768c9768 smb/server: do not require
delete access for non-replacing links |
Author: ChenXiaoSong [email protected] Upstream:
851ed9e09639e0daf79a506ce26097b296ed5518 File changed:
fs/smb/server/smb2pdu.c (8 insertions, 6 deletions)
This is surprising — it's a ksmbd (SMB server) patch for hard link
permissions, yet it breaks all input devices (not just HID++/Logitech
Unifying, but also generic USB and PS/2 keyboards). The system becomes
completely unusable as no keyboard input is recognized, not even at the
GRUB level on subsequent boots.
Additional finding: nvidia-independent
As requested, I tested 6.12.100 with
|module_blacklist=nvidia,nvidia_modeset,nvidia_drm,nvidia_uvm| on the
kernel command line. The input device failure persists identically
without any proprietary modules loaded. This is a pure kernel
regression, independent of nvidia.
Bisect log
|git bisect start # good: [6d15a1029d425b15c59463910ebdccc4afe760d6]
Linux 6.12.96 git bisect good 6d15a1029d425b15c59463910ebdccc4afe760d6 #
bad: [52a355b23cd2dd77b683ed1a1bbe2b7408bd9c74] Linux 6.12.100 git
bisect bad 52a355b23cd2dd77b683ed1a1bbe2b7408bd9c74 # bad:
[91cacfb19bbcf11f81f2269163aa0b56655c3ae7] net: dsa: realtek: fix memory
leak in rtl8366rb_setup_led() git bisect bad
91cacfb19bbcf11f81f2269163aa0b56655c3ae7 # bad:
[a9f8a1d2e3ff511eafd4c5462481950c2f4d2b5d] vhost/net: complete zerocopy
ubufs only once git bisect bad a9f8a1d2e3ff511eafd4c5462481950c2f4d2b5d
# bad: [a7a65cc2463786d135a28f8bf2c658e37ba64617] RDMA/mana_ib: Use
ib_get_eth_speed for reporting port speed git bisect bad
a7a65cc2463786d135a28f8bf2c658e37ba64617 # bad:
[6f8c8372e80e3eab24ea9bdcd42cb6636cbc57a8] dt-bindings: pinctrl:
nvidia,tegra234: Add missing required block git bisect bad
6f8c8372e80e3eab24ea9bdcd42cb6636cbc57a8 # bad:
[bf9b5bb0a149e4a37f257dfeecac23e97339c802] crypto: hisi-trng - Remove
crypto_rng interface git bisect bad
bf9b5bb0a149e4a37f257dfeecac23e97339c802 # bad:
[d306da8833e75f669d93424fd84940236f3850bc] nvmet-tcp: check INIT_FAILED
before nvmet_req_uninit in digest error path git bisect bad
d306da8833e75f669d93424fd84940236f3850bc # bad:
[307010f4693eac9401b5be1c7bbf58ac65876310] bpf: Prefer dirty packs for
eBPF allocations git bisect bad 307010f4693eac9401b5be1c7bbf58ac65876310
# bad: [eed774da601268dae674e14d54a15e3624691f52] bpf: Support for
hardening against JIT spraying git bisect bad
eed774da601268dae674e14d54a15e3624691f52 # bad:
[c716a59e9977d751e5eb54bcfa6a80124cb5067b] iommu/vt-d: Clear Present bit
before tearing down context entry git bisect bad
c716a59e9977d751e5eb54bcfa6a80124cb5067b # bad:
[dcc1bce68aec19333ed84d7f315a67e6768c9768] smb/server: do not require
delete access for non-replacing links git bisect bad
dcc1bce68aec19333ed84d7f315a67e6768c9768 # first bad commit:
[dcc1bce68aec19333ed84d7f315a67e6768c9768] smb/server: do not require
delete access for non-replacing links |
Hardware (for reference)
* Motherboard: Gigabyte X99-UD5 WIFI (BIOS F24c)
* CPU: Intel Xeon (Haswell-E, 14C/28T)
* RAM: 96GB (6/8 slots)
* GPU: NVIDIA GeForce GTX 1080 Ti (GP102)
* Input: Logitech ERGO K860 + MX Vertical via Unifying receiver
(046d:c52b)
* Also tested: generic USB keyboard and PS/2 keyboard — all fail
* Desktop: KDE Plasma 6.3.6 / Wayland / PipeWire 1.4.2
Note
Given that the first bad commit is an SMB/ksmbd patch that should have
no relation to input devices, this might indicate a build/patch ordering
issue in the 6.12.100 stable release rather than the SMB patch itself
being the culprit. All bisect kernels were built using |localmodconfig|
from the running 6.12.96+deb13-amd64 configuration.
Best regards, Salvador García ServiTech - Servicios y Tecnología
Computacional
En 01/08/26 14:07, Salvatore Bonaccorso escribió:
Control: tags -1 + moreinfo
Hi,
On Sat, Aug 01, 2026 at 07:28:07PM +0000, Salvador Rafael García Pliego! wrote:
Package: linux-image-6.12.100+deb13-amd64
Version: 6.12.100-1
Severity: grave
Tags: regression
Regression: systemd-modules-load.service failure causes HID++ devices to
fall back to hid-generic, resulting in broken input devices
Debian Bug Report: linux-image-6.12.100+deb13-amd64
Subject
Regression: systemd-modules-load.service failure causes HID++ devices to
fall back to hid-generic, resulting in broken input devices and ghost
keyboard events
Package
linux-image-6.12.100+deb13-amd64
Version
6.12.100-1
Severity
grave (renders system unusable — keyboard/mouse non-functional)
Tags
regression
------------------------------------------------------------------------
Description
After upgrading from kernel 6.12.96+deb13-amd64 to 6.12.100+deb13-amd64, the
system becomes effectively unusable due to broken HID input device
initialization. The Logitech Unifying receiver (046d:c52b) fails to
initialize with the specialized |hid_logitech_dj| and |hid_logitech_hidpp|
drivers, falling back to |hid-generic|. This results in:
1. Keyboard (Logitech ERGO K860) largely non-functional
2. Mouse (Logitech MX Vertical) degraded functionality
3. Ghost keyboard events (spurious KeyRelease and KeymapNotify events)
4. Multimedia keys (volume, play/pause, track skip) completely
non-functional
5. Volume OSD appearing spontaneously without user input
Reverting to 6.12.96+deb13-amd64 immediately resolves all issues.
Root Cause Analysis
The |systemd-modules-load.service| fails during boot with kernel 6.12.100
because the NVIDIA kernel modules (|nvidia-current|,
|nvidia-current-modeset|, |nvidia-current-drm|) are not found in
|/lib/modules/6.12.100+deb13-amd64/|. This service failure appears to
cascade, preventing or delaying the loading of other kernel modules
including |hid_logitech_hidpp| and |hid_logitech_dj|.
Evidence: Boot with 6.12.100 (BROKEN)
The Logitech Unifying receiver is detected only with the generic HID driver:
|kernel: input: Logitech USB Receiver as .../input/input13 kernel:
hid-generic 0003:046D:C52B.0003: input,hidraw2: USB HID v1.11 Keyboard
[Logitech USB Receiver] on usb-0000:00:14.0-12/input0 kernel: input:
Logitech USB Receiver Mouse as .../input/input14 kernel: input: Logitech USB
Receiver Consumer Control as .../input/input15 kernel: input: Logitech USB
Receiver System Control as .../input/input16 kernel: hid-generic
0003:046D:C52B.0004: input,hiddev2,hidraw3: USB HID v1.11 Mouse [Logitech
USB Receiver] on usb-0000:00:14.0-12/input1 kernel: hid-generic
0003:046D:C52B.0005: hiddev3,hidraw4: USB HID v1.11 Device [Logitech USB
Receiver] on usb-0000:00:14.0-12/input2|
Note: Individual devices (MX Vertical, ERGO K860) are *never enumerated by
name*. The |logitech-djreceiver| and |logitech-hidpp-device| drivers are
never invoked.
Evidence: Boot with 6.12.96 (WORKING)
The specialized Logitech drivers load correctly:
|kernel: logitech-djreceiver 0003:046D:C52B.0005: hiddev2,hidraw2: USB HID
v1.11 Device [Logitech USB Receiver] on usb-0000:00:14.0-12/input2 kernel:
input: Logitech MX Vertical as .../input/input32 kernel:
logitech-hidpp-device 0003:046D:407B.0006: input,hidraw3: USB HID v1.11
Keyboard [Logitech MX Vertical] on usb-0000:00:14.0-12/input2:1 kernel:
input: Logitech ERGO K860 as .../input/input33 kernel: logitech-hidpp-device
0003:046D:4088.0007: input,hidraw4: USB HID v1.11 Keyboard [Logitech ERGO
K860] on usb-0000:00:14.0-12/input2:2 kernel: logitech-hidpp-device
0003:046D:4088.0007: HID++ 4.5 device connected.|
systemd-modules-load.service failure log (6.12.100)
|systemd-modules-load: modprobe: FATAL: Module nvidia-current not found in
directory /lib/modules/6.12.100+deb13-amd64 systemd-modules-load: modprobe:
ERROR: could not insert 'nvidia': Invalid argument systemd-modules-load:
modprobe: FATAL: Module nvidia-current-modeset not found in directory
/lib/modules/6.12.100+deb13-amd64 systemd-modules-load: modprobe: ERROR:
could not insert 'nvidia_modeset': Invalid argument systemd-modules-load:
modprobe: FATAL: Module nvidia-current-drm not found in directory
/lib/modules/6.12.100+deb13-amd64 systemd-modules-load: Failed to insert
module 'nvidia_drm': Invalid argument systemd-modules-load: Failed to find
module 'vmmon' systemd-modules-load: Failed to find module 'vmnet'
systemd[1]: systemd-modules-load.service: Main process exited, code=exited,
status=1/FAILURE systemd[1]: Failed to start systemd-modules-load.service -
Load Kernel Modules.|
Additional kernel errors observed with 6.12.100
|kernel: pci 0000:00:1c.0: retraining failed kernel: EDAC sbridge: Failed to
register device with error -19 kernel: ACPI BIOS Error (bug): Could not
resolve symbol [\_SB.PCI0.SAT1.SPT0._GTF.DSSP], AE_NOT_FOUND kernel: ACPI
Error: Aborting method \_SB.PCI0.SAT1.SPT0._GTF due to previous error
(AE_NOT_FOUND)|
Observed symptoms in desktop session (KDE Plasma 6.3.6 / Wayland)
When the system managed to boot (with degraded input), the following was
observed via |xev|:
* Spurious |KeyRelease| events for keys not pressed (e.g., |v|, |Return|)
* Repeated |KeymapNotify| events with |keys: 130 0 0 0...| (bits 1 and
7 of byte 0 stuck)
* These events were NOT captured by |libinput debug-events|,
confirming they originated from the generic HID layer, not from
proper device enumeration
* Volume OSD appeared spontaneously; |pactl subscribe| confirmed real
sink volume changes on sink #67
* Multimedia keys (volume, play/pause, track navigation) either
non-functional or firing once then freezing
Hardware
* *Motherboard:* Gigabyte X99-UD5 WIFI (BIOS F24c, 2018-06-15)
* *CPU:* Intel (Haswell-E/EP, 2399 MHz)
* *RAM:* 96GB (6/8 slots populated)
* *GPU:* NVIDIA GeForce GTX 1080 Ti (GP102)
* *Keyboard:* Logitech ERGO K860 (PID: 4088) via Unifying receiver
* *Mouse:* Logitech MX Vertical (PID: 407b) via Unifying receiver
* *Unifying Receiver:* Logitech (046d:c52b) on USB 3-12
* *Audio:* Realtek ALC1150 (analog surround 7.1)
* *Webcam:* Microsoft LifeCam Studio
* *WiFi:* Intel (iwlwifi)
* *Desktop:* KDE Plasma 6.3.6 on Wayland
* *Audio server:* PipeWire 1.4.2
Steps to Reproduce
1. Boot system with kernel 6.12.96+deb13-amd64 — all input devices work
correctly
2. Install and boot kernel 6.12.100+deb13-amd64
3. Observe: keyboard/mouse via Logitech Unifying receiver are broken or
severely degraded
4. Revert to 6.12.96 — all issues resolved
Workaround
Boot with kernel 6.12.96+deb13-amd64 and hold the 6.12.100 package:
|sudo apt-mark hold linux-image-6.12.100+deb13-amd64 sudo apt-mark hold
linux-image-amd64|
Additional Notes
Multiple rapid reboots were attempted with 6.12.100 (5+ boot attempts within
~10 minutes on 2026-07-31, visible in |journalctl --list-boots|), none of
which resolved the issue. The problem is 100% reproducible on every boot
with the affected kernel.
The ACPI errors and |LPM support broken| messages on SATA ports are also
present but may be pre-existing on this X99 platform and not directly
related to the HID regression. They are included for completeness.
Sad to hear about your regression between 6.12.96 and 6.12.100.
Can you please bisect the issue with the following procedure so we
know what is causing your failure:
git clone --single-branch -b
linux-6.12.yhttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
cd linux-stable
git checkout v6.12.96
cp /boot/config-$(uname -r) .config
yes '' | make localmodconfig
make savedefconfig
mv defconfig arch/x86/configs/my_defconfig
# test 6.12.96 to ensure this is "good"
make my_defconfig
make -j $(nproc) bindeb-pkg
... install the resulting .deb package and confirm problem does not exist
# test 6.12.100 to ensure this is "bad"
git checkout v6.12.100
make my_defconfig
make -j $(nproc) bindeb-pkg
... install the resulting .deb package and confirm problem exists
With that confirmed, the bisection can start:
git bisect start
git bisect good v6.12.96
git bisect bad v6.12.100
In each bisection step git checks out a state between the oldest
known-bad and the newest known-good commit. In each step test using:
make my_defconfig
make -j $(nproc) bindeb-pkg
... install, verify if problem exists
and if the problem is hit run:
git bisect bad
and if the problem doesn't trigger run:
git bisect good
. Please pay attention to always select the just built kernel for
booting, it won't always be the default kernel picked up by grub.
Iterate until git announces to have identified the first bad commit.
Then provide the output of
git bisect log
In the course of the bisection you might have to uninstall previous
kernels again to not exhaust the disk space in /boot. Also in the end
uninstall all self-built kernels again.
Additional note: Assuming your HID regression is independent, but you
have proprietary nvidia modules loaded (or not with 6.12.100), in
particular please thus try to reproduce the issue first in v6.12.100
withouth proprietary modules loading (which we need to have a untained
kernel for upstream reporting later).
Thanks for your time in helping to debug this new issue!
Regards,
Salvatore
