New: rtl8822cu: BLE disconnects with firmware 0x0673_0EDC regression

https://bugzilla.kernel.org/show_bug.cgi?id=221309

            Bug ID: 221309
           Summary: rtl8822cu: BLE disconnects with firmware 0x0673_0EDC
                    regression
           Product: Drivers
           Version: 2.5
          Hardware: AMD
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P3
         Component: Bluetooth
          Assignee: [email protected]
          Reporter: [email protected]
        Regression: No

Created attachment 309794
--> https://bugzilla.kernel.org/attachment.cgi?id=309794&action=edit
stability log after downgrade

Overview

After updating linux-firmware-realtek to 20260221-1 (which includes commit fbf93476,
"update rtl8822c bt usb and uart firmware to 0x0673"), the RTL8822CU USB Bluetooth adapter
(0bda:b00c) begins dropping BLE connections every few minutes. Each disconnect lasts several seconds before the device reconnects. The adapter itself never resets and
the USB device remains active throughout. Only the BLE link is lost.

Steps to Reproduce

1. Use a system with an RTL8822CU USB Bluetooth adapter (0bda:b00c).
2. Pair and connect a BLE HID device (tested: Logitech M585/M590 mouse).
3. Optionally connect a second BLE HID device simultaneously (tested:
HHKB-Hybrid_2 keyboard).
4. Use the BLE device normally. No special interaction required, disconnects occur at idle and
during active use.

Actual Results

The BLE mouse disconnects often (observed 13+ times over 60 minutes). bluetoothd reports
Connected: false on the device D-Bus object, the kernel destroys and recreates the UHID device
on reconnect, and the mouse is unresponsive for the duration. Downtime can be multiple seconds.
No kernel messages are produced during the disconnect. The second BLE device (HHKB keyboard)
is unaffected.

dbus-monitor confirms the disconnect is hardware-initiated.

Expected Results

The BLE connection should remain stable during normal use, as it did with the previous firmware.

Build Date & Hardware

Offending firmware: rtl_bt/rtl8822cu_fw.bin, version 0x0673_0EDC, introduced
in commit

    fbf93476b78933a87073d3bf116a84f2599b37b5 (2026-02-02)

Adapter: Realtek RTL8822CU, USB ID 0bda:b00c
Kernel: 6.19.10-arch1-1
bluetoothd: 5.86
OS: Arch Linux

Additional Builds and Platforms

Does not occur with rtl_bt/rtl8822cu_fw.bin from tag 20260110 (the commit immediately
preceding fbf93476). Downgrading to that firmware eliminates all disconnects.

Additional Information

USB autosuspend was ruled out by polling /sys/bus/usb/devices/1-3/power/runtime_status and runtime_suspended_time during a live disconnect. The adapter remained active throughout.
The disconnect also happens while actively moving the mouse. Everything tested with default Bluetooth settings.

I'm currently downgrading linux-firmware-realtek to 20260110-1 as a workaround.

I'm running two linux user accounts in parallel, one at tty1 and one at tty2.

You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.

https://bugzilla.kernel.org/show_bug.cgi?id=221309

[email protected] changed:

What |Removed |Added
----------------------------------------------------------------------------

            Summary|rtl8822cu: BLE disconnects  |[REGRESSION] rtl8822cu: BLE
                   |with firmware 0x0673_0EDC   |disconnects with firmware
                   |regression                  |0x0673_0EDC

You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.

https://bugzilla.kernel.org/show_bug.cgi?id=221309

[email protected] changed:

What |Removed |Added
----------------------------------------------------------------------------

 Bisected commit-id|                            |fbf93476
     Kernel Version|                            |6.19.10
         Regression|No                          |Yes

You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.

[To/Cc: +Hilda, linux-firmware@, Josh]

Dear Hilda, dear Josh,

Hilda, unfortunately, you do not have a Linux Kernel Bugzilla account, so I am replying here.

Am 02.04.26 um 01:57 schrieb [email protected]:

[… Linux 6.19.10-arch1-1 …]

It looks like, the linux-firmware commit [1][2] caused a regression. (Also linux-bluetooth@ wasn’t in Cc?) Could you please look into it? I recommend to revert it for the time being and cut a new linux-firmware release.

Kind regards,

Paul

PS: Josh, do you have a pointer how to reach the commit from the merge request?

https://gitlab.com/kernel-firmware/linux-firmware/-/merge_requests/900 shows https://gitlab.com/kernel-firmware/linux-firmware/-/commit/1f806c75e479fea9f222579deb6e69d285b614fc, but I do not see a way to get to https://gitlab.com/kernel-firmware/linux-firmware/-/commit/fbf93476.

[1]: https://lore.kernel.org/all/[email protected]/
[2]: https://gitlab.com/kernel-firmware/linux-firmware/-/commit/fbf93476

https://bugzilla.kernel.org/show_bug.cgi?id=221309

--- Comment #1 from Paul Menzel ([email protected]) --- Thank you for your detailed report. As Hilda does not seem to have Linux Kernel Bugzilla account, I replied to the Bugzilla notification on the [Bluetooth mailing list suggesting to revert the commit][1]. I wasn’t sure, if you would be ok with me putting your email address in Cc making it public. It would probably ease things up, if you would be in the loop.

[1]:
https://lore.kernel.org/linux-bluetooth/[email protected]%2F/T/#u

You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.

https://bugzilla.kernel.org/show_bug.cgi?id=221309

--- Comment #2 from [email protected] ---
Yes, absolutely OK. If I can help diagnose this further I'll gladly help. It's not easy for me to debug this by myself.

You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.

Hi Paul,

Thank you for notifying us and providing the detailed information.
We have already reverted the commit. We will proceed with fixing this issue and provide an updated firmware once the fix is complete. Thank you again.

Regards,
Hilda

-----Original Message-----
From: Paul Menzel <[email protected]>
Sent: Thursday, April 2, 2026 6:22 PM
To: Hilda Wu <[email protected]>
Cc: [email protected]; [email protected]; Josh Boyer <[email protected]>; [email protected]
Subject: Re: [REGRESSION] rtl8822cu: BLE disconnects with firmware 0x0673_0EDC

External mail : This email originated from outside the organization. Do not reply, click links, or open attachments unless you recognize the sender and know the content is safe.

[To/Cc: +Hilda, linux-firmware@, Josh]

Dear Hilda, dear Josh,

Hilda, unfortunately, you do not have a Linux Kernel Bugzilla account, so I am replying here.

Am 02.04.26 um 01:57 schrieb [email protected]:

[... Linux 6.19.10-arch1-1 ...]

It looks like, the linux-firmware commit [1][2] caused a regression.
(Also linux-bluetooth@ wasn't in Cc?) Could you please look into it? I recommend to revert it for the time being and cut a new linux-firmware release.

Kind regards,

Paul

PS: Josh, do you have a pointer how to reach the commit from the merge request?

https://gitlab.com/kernel-firmware/linux-firmware/-/merge_requests/900
shows
https://gitlab.com/kernel-firmware/linux-firmware/-/commit/1f806c75e479fea9f222579deb6e69d285b614fc,
but I do not see a way to get to
https://gitlab.com/kernel-firmware/linux-firmware/-/commit/fbf93476.

[1]:
https://lore.kernel.org/all/[email protected]/
[2]: https://gitlab.com/kernel-firmware/linux-firmware/-/commit/fbf93476

https://bugzilla.kernel.org/show_bug.cgi?id=221309

--- Comment #3 from Paul Menzel ([email protected]) --- Hilda replied [1]:

[1]:
https://lore.kernel.org/linux-bluetooth/[email protected]%2F/T/#me79bb1b45c47cbda2de692a6499ecd514709c0e2

You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.