RE: [BUG] rtw89: RTL8852CE RX_DCK timeout + Bluetooth A2DP coexistence failures (fw-2.bin 0.27.129.4)

[Ping-Ke Shih <pkshih@realtek.com>]

or1k <tixzit@gmail.com> wrote:
> Hi,
> 
> Reporting a Wi-Fi/Bluetooth coexistence issue on RTL8852CE that also has an
> open Fedora/Red Hat tracker (bugzilla.redhat.com #2349675), but that report
> was filed against an older firmware (rtw8852c_fw-1.bin, 0.27.97.0). I'm
> seeing what looks like the same class of failure on the newer
> rtw8852c_fw-2.bin, firmware version 0.27.129.4 (3f1a5302), so it may not be
> fully fixed by whatever changed between those firmware revisions.

We are working on driver to support the new firmware (0.27.129.4), and 
I expected to land the patches on 7.3. 

I'd clarify that does 0.27.97.0 work to you?

> 
> Hardware:
>   Wi-Fi:      Realtek RTL8852CE [10ec:c852] rev 01, Lenovo subsystem
> [17aa:5852]
>               PCI 0000:02:00.0, driver rtw89_8852ce
>   Bluetooth:  same combo chip, USB 0bda:5852 "Realtek Bluetooth Radio",
>               driver btusb/btrtl, BT firmware rtl8852cu_fw_v2.bin (0x040d7225)
>   Platform:   Lenovo laptop, AMD Ryzen APU (unrelated to this bug, just for
>               context — ruled out as a factor)
> 
> Software:
>   Kernel 7.1.3-200.fc44.x86_64 (Fedora 44), linux-firmware-20260622-1.fc44
> 
> Symptom: intermittent Wi-Fi disconnects/reauth together with Bluetooth A2DP
> audio transport failures, both traced to the same physical combo chip.
> 
> dmesg (Wi-Fi side), recurring several times/day in short clusters:
> 
>   rtw89_8852ce 0000:02:00.0: [RX_DCK] S1 RXDCK timeout
>   rtw89_8852ce 0000:02:00.0: timed out to flush queues
> 
> journalctl (Bluetooth side), same time frame, device 85:42:36:51:B1:D1
> (Bluetooth headphones, A2DP sink):
> 
>   wireplumber: spa.bluez5.sink.media: Misscket
>     (0 suppressed): Bluetooth adapter firm
>   bluetoothd: src/profile.c:ext_io_disconn for
>     Hands-Free Voice gateway: getpeername:
>     connected (107)
>   wireplumber: spa.bluez5: Failure in Bluetooth audio transport
>     /org/bluez/hci0/dev_85_42_36_51_B1_D1/
> 
> Direct correlation captured with mtr: a WireGuard-tunneled ping test
> (ultimately routed over wlp2s0) showed 5.9 the
> same ~60s window as 8 consecutive "Missingines
> for the same device (00:12:51) and an HFP 1).
> Both radios degraded together, which pointmware
> fault rather than two unrelated bugs.
> 
> Workarounds tried, partial effect only:
>   - USB autosuspend disabled for the BT ad
>   - rtw89_core.disable_ps_mode=1, rtw89_pc
>   - A2DP codec forced to SBC instead of AA
> 
> Is this expected to be covered by the same fix path as RH#2349675, or is it
> worth tracking separately since it reprodu?

Can you explain what RH#2349675 is?

> Happy to test candidate firmware/patches o

I will share the patches when they get ready (I suppose it will be
in one or two weeks).

Ping-Ke