All of lore.kernel.org
 help / color / mirror / Atom feed
From: Bitterblue Smith <rtl8821cerfe2@gmail.com>
To: Daniel <dani069xp@gmail.com>, linux-wireless@vger.kernel.org
Cc: pkshih@realtek.com
Subject: Re: [Bug report] rtw89: 8851be: WiFi disconnect (reason=4, beacon loss) 1-2 s after A2DP stream start on the card's own Bluetooth, 5 GHz 80 MHz
Date: Fri, 20 Nov 2026 22:11:18 +0200	[thread overview]
Message-ID: <228ba38a-e5a2-4d8d-9547-788b9f0aec8a@gmail.com> (raw)
In-Reply-To: <CALovXdg6sOyK9orNrXSRSoSkVf7Adqa8xNC8Q=ESw4f0YEryaQ@mail.gmail.com>

On 20/11/2026 21:58, Daniel wrote:
> Hi,
> 
> On an RTL8851BE combo card, starting an A2DP audio stream on the card's own
> Bluetooth reliably makes the WiFi link drop within 1-2 seconds. The station is
> disconnected locally with reason=4 and the AP then ignores re-authentication
> for 50-90 s.
> 
> Hardware / software
> - WiFi: Realtek RTL8851BE PCIe 802.11ax [10ec:b851], driver rtw89_8851be
> - Bluetooth: USB 0bda:b850 (btusb, rtl_bt/rtl8851bu_fw.bin) on the same card
> - Firmware: rtw89/rtw8851b_fw-1.bin, version 0.29.41.5 (4bd6ebac)
> - linux-firmware 20261019
> - Kernel 7.3.5-zen1-1-zen (also seen on 7.3.4-zen2-1 and zen2-2), Arch Linux
> - NetworkManager 1.58.1, wpa_supplicant 2.12, bluez 5.88, PipeWire 1.6.11
> - Module options: rtw89_core disable_ps_mode=y,
>   rtw89_pci disable_aspm_l1=y disable_aspm_l1ss=y disable_clkreq=y (no effect)
> - NM: wifi.powersave=2, MAC randomization "stable"
> - AP: 5 GHz, channel 157, 80 MHz VHT; signal about -58 to -61 dBm, 1x1
> - BT audio: A2DP SBC to a headset (sink)
> 
> Observed
> bluetoothd logs "fd(N) ready" when the A2DP transport starts. In every case
> wpa_supplicant reports CTRL-EVENT-DISCONNECTED reason=4 locally_generated=1
> within 1-2 s:
> 
>   2026-11-14 13:08:32 BT stream  -> 13:08:34 WiFi drop (12 more drops
> while streaming)
>   2026-11-20 13:06:56 BT connect -> 13:07:04 WiFi drop; another at 13:47:54
>   2026-11-20 13:53:37 BT stream  -> 13:53:38 WiFi drop, reconnected 13:54:28
> 
> "iw dev wlp4s0 station dump" shows the beacon loss counter increasing with each
> event. ICMP loss to the AP is 0% outside these events and TX retries/failures
> are 0. Other devices on the same AP are not affected.
> 
> Width dependence
> - 80 MHz: 4 of 4 stream starts dropped the link.
> - 40 MHz (same AP, kernel, signal): 1 drop in 3 stream starts (2 were clean
>   over several days of use).
> - With Bluetooth off, 80 MHz showed no drops in the short test afterwards.
>   A USB-tethered phone on the same machine did not show the problem.
> 
> Not the cause (tested): browser/GPU, NetworkManager config, MAC randomization,
> power save, ASPM options, kernel 7.3.4 -> 7.3.5.
> 
> Expected: a stable WiFi link while Bluetooth A2DP streams, as it does on the
> same card with a narrower channel.
> 
> Coex trace around one drop (debugfs btc_info sampled every 0.5 s, local time,
> 5 GHz ch155 center, bw=2, 80 MHz; "iw event -t" in parallel)
> 
>   14:20:54.975  BT connect:Y (A2DP_active:Y from 14:20:55.5)
>   14:20:57.994  BT low-priority requests jump: Lo-rx/Lo-tx 0-3 -> 3064
>   14:20:59.504  BT profile:A2DP (bluetoothd: transport fd ready at 14:20:59)
>   14:21:00.830  iw event: CQM event: beacon loss
>   14:21:00.958  iw event: deauth reason 4 "Disassociated due to inactivity"
>                 (disconnected, local request)
>   14:21:01.020  coex: WL connect=NO_LINK, reason NTFY_POWEROFF, action WL_OFF
> 
> Second, independent capture of the same event (same setup, raw files kept):
> 
>   14:49:31.4    BT connect:Y
>   14:49:34.434  BT Lo-rx/Lo-tx requests jump 0-239 -> 3229
>   14:49:35      bluetoothd: A2DP transport fd ready
>   14:49:35.945  BT profile:A2DP
>   14:49:36.797  iw event: CQM event: beacon loss
>   14:49:36.910  iw event: deauth reason 4 "Disassociated due to inactivity"
>   14:49:36.952  coex: WL NO_LINK, action WL_OFF
>   14:50:13      WiFi connected again (about 36 s; "auth: timed out" x4 before)
> 
> In both captures the coex state is action:WL_5G, ant_path:FDD, tdma type:0
> right up to the last sample before the drop.
> 
> From 14:20:50 up to the drop the coex manager stays in action:WL_5G,
> ant_path:FDD, tdma type:0 the whole time, i.e. it does not arbitrate WL/BT
> while the BT traffic ramps up. hw_info reports ant_iso:10, rfe_type:0x2,
> tx_num:1, rx_num:1. WL_FW_coex 7.9.0 (desired 6.0.0, "Match"), BT_FW_coex 7.
> rssi was -54 dBm at the last sample before the drop (VHT MCS 9, 80 MHz).
> 
> For comparison, when the same station is on 2.4 GHz (ch 6, 20 MHz), A2DP
> streaming works for me without drops (user observation, not traced).
> The failing case is only 5 GHz + BT A2DP, and it is worse with 80 MHz than
> with 40 MHz.
> 
> Full btc_info trace, ser_counters (all 0) and the iw event log are available;
> I can send them or try patches/debug_mask settings. Thanks for any pointers.
> 
> Thanks,
> Dani

Hi,

The fix for this will be in kernel 7.4-rc4, and I think it will be added
automatically to 7.3 as well at some point:

https://lore.kernel.org/linux-wireless/E6568F470624C07E+179429427543.116514.14355630958613033614@guwolf.com/

  reply	other threads:[~2026-11-20 20:11 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-20 19:58 [Bug report] rtw89: 8851be: WiFi disconnect (reason=4, beacon loss) 1-2 s after A2DP stream start on the card's own Bluetooth, 5 GHz 80 MHz Daniel
2026-11-20 20:11 ` Bitterblue Smith [this message]
2026-11-20 20:20   ` Daniel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=228ba38a-e5a2-4d8d-9547-788b9f0aec8a@gmail.com \
    --to=rtl8821cerfe2@gmail.com \
    --cc=dani069xp@gmail.com \
    --cc=linux-wireless@vger.kernel.org \
    --cc=pkshih@realtek.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.