[BUG] WARNING: ODEBUG bug in corsair_remove (drivers/hid/hid-corsair.c, v7.4-rc3)
From: Zheng Zheng
Date: Mon Nov 16 2026 - 22:44:16 EST
Hi Jiri, Benjamin,
We found a reproducible debugobjects warning in the Corsair HID driver on
v7.4-rc3 with syzkaller. We reproduced this warning on an unmodified
v7.4-rc3 kernel.
Key log excerpt from the rc3 external run
(logs/external-verification/run1-serial.log, hit at 93.23 s):
[ 93.233905][ T11] ODEBUG: free active (active state 0) object:
ff1100017cebb2c8 object type: work_struct hint:
k90_record_led_work+0x0/0x210
[ 93.233944][ T11] WARNING: lib/debugobjects.c:629 at
debug_print_object+0x1a0/0x2e0, CPU#0: kworker/0:1/11
[ 93.237255][ T11] Modules linked in: dm_multipath(E)
[ 93.238355][ T11] CPU: 0 UID: 0 PID: 11 Comm: kworker/0:1
Tainted: G E 7.4.0-rc3-kernsight-kasan #1
PREEMPT(full)
[ 93.242242][ T11] Workqueue: usb_hub_wq hub_event
[ 93.242850][ T11] RIP: 0010:debug_print_object+0x1a5/0x2e0
[ 93.258269][ T11] Call Trace:
[ 93.258659][ T11] <TASK>
[ 93.259006][ T11] ? __pfx_k90_record_led_work+0x10/0x10
[ 93.259685][ T11] debug_check_no_obj_freed+0x3d7/0x520
[ 93.262901][ T11] kfree+0x3ff/0x660
[ 93.263373][ T11] ? corsair_remove+0xff/0x2f0
[ 93.263953][ T11] corsair_remove+0xff/0x2f0
[ 93.265130][ T11] hid_device_remove+0xd6/0x200
[ 93.266372][ T11] device_remove+0xd0/0x170
[ 93.266925][ T11] device_release_driver_internal+0x45f/0x630
[ 93.268915][ T11] bus_remove_device+0x2bf/0x570
[ 93.270799][ T11] device_del+0x39b/0x9e0
[ 93.273129][ T11] hid_destroy_device+0x1a9/0x240
[ 93.273736][ T11] usbhid_disconnect+0xa8/0xe0
[ 93.274289][ T11] usb_unbind_interface+0x1ea/0x9a0
[ 93.275595][ T11] device_remove+0x12a/0x170
[ 93.276150][ T11] device_release_driver_internal+0x45f/0x630
[ 93.278564][ T11] bus_remove_device+0x2bf/0x570
[ 93.279884][ T11] device_del+0x39b/0x9e0
[ 93.281551][ T11] usb_disable_device+0x35e/0x800
[ 93.282146][ T11] usb_disconnect+0x2e4/0xa10
[ 93.282725][ T11] hub_event+0x1ea9/0x5010
[ 93.294782][ T11] Kernel panic - not syncing: kernel:
panic_on_warn set ...
Critical call chain:
debug_check_no_obj_freed
kfree
corsair_remove+0xff/0x2f0
hid_device_remove
device_remove
device_release_driver_internal
bus_remove_device
device_del
hid_destroy_device
usbhid_disconnect
usb_unbind_interface
usb_disable_device
usb_disconnect
hub_event
The ODEBUG hint is `k90_record_led_work`.
How to reproduce:
1. Build vanilla v7.4-rc3 with:
CONFIG_KASAN=y
CONFIG_DEBUG_OBJECTS=y
CONFIG_DEBUG_OBJECTS_WORK=y
CONFIG_HID_CORSAIR=y
CONFIG_DEBUG_INFO=y
The bzImage used here has sha256
0c16409f8aec508ac44a63e631c2bd352536401f3fd0856df1d839fb953a1882.
2. Boot a Debian trixie x86_64 guest (6 GiB RAM, 4 vCPU, KVM) with at least:
root=/dev/sda console=ttyS0 oops=panic panic_on_warn=1
The full QEMU command line is in environment.txt.
3. Copy syz-execprog and syz-executor built from syzkaller commit
853b104bb454141a5b0fc0e4218ec78e9f34ffb1, plus
reproducer/repro.kernsight.prog, into the guest.
4. In the guest, as root, run:
syz-execprog -procs=8 -threaded -repeat=1 \
-executor ./syz-executor repro.kernsight.prog
Expected result: ODEBUG `free active` warning in dmesg/serial after roughly
90-100 seconds, with the call chain above; panic_on_warn=1 then panics the
guest. We observed the hit 2/2 on fresh VMs:
Run 1: 93.23 s after boot
Run 2: 94.27 s after boot
Analysis:
The Corsair LED work items are queued by k90_brightness_set() via
schedule_work(). During removal, both cleanup paths do:
cancel_work_sync(&...->work);
led_classdev_unregister(&...->cdev);
kfree(...);
led_classdev_unregister() turns the LED off, which can queue the work item
again after the first cancel_work_sync(). The containing structure is then
freed while the work_struct is still active, so debugobjects reports
"free active" in corsair_remove().
A reproducer repository is available at:
https://github.com/Zheng-zheng-codes/corsair-remove-odebug-warning
It contains the syzkaller program, kernel config, guest/run details and the
full rc3 serial logs.
Best regards,
Zheng Zheng