[PATCH v2] wifi: ath11k: apply existing PM quirk to ThinkPad P14s Gen 5 AMD

Fri May 1 00:08:19 PDT 2026

On Mon, Apr 13, 2026 at 11:43 PM Baochen Qiang
<baochen.qiang at oss.qualcomm.com> wrote:
>
>
>
> On 4/1/2026 11:48 AM, Kyle Farnung wrote:
> > On Tue, Mar 31, 2026 at 7:08 PM Baochen Qiang
> > <baochen.qiang at oss.qualcomm.com> wrote:
> >>
> >>
> >>
> >> On 3/31/2026 2:32 PM, Kyle Farnung via B4 Relay wrote:
> >>> From: Kyle Farnung <kfarnung at gmail.com>
> >>>
> >>> Some ThinkPad P14s Gen 5 AMD systems experience suspend/resume
> >>> reliability issues similar to those reported in [1]. These platforms
> >>
> >> how similar it is? can you describe the issue in details?
> >
> > The issue is that intermittently after suspend my WiFi adapter connects
> > successfully for a few minutes and then drops. It will then keep trying to
> > reconnect in a loop but never succeed. A reboot will fix it, but eventually
> > I found that reloading the module also resolves the issue
> > (modprobe -r ath11k_pci && modprobe ath11k_pci). Based on some searching, I
> > did try adding "ath11k_pci.disable_idle_ps=1" to my kernel arguments. At
> > first it looked like maybe it worked, but then I hit the same problem
> > again. At that point I decided to try building a custom module with the
> > ATH11K_PM_WOW override and so far I'm two days and 10 suspends in without
> > issue.
> >
> > Looking through kernel logs, the issue appears to have started with kernel
> > version 6.17.4. It looks like my Fedora install jumped from 6.16.10 to
> > 6.17.4 on October 22, 2025 and I started seeing the issue two days later.
> >
> > Here are the logs from the most recent occurrence (filtered for brevity):
> >
> > Mar 29 15:26:24 kjfp14sg5 kernel: PM: suspend exit
> > Mar 29 15:26:24 kjfp14sg5 kernel: ath11k_pci 0000:02:00.0: chip_id
> > 0x12 chip_family 0xb board_id 0xff soc_id 0x400c1211
> > Mar 29 15:26:24 kjfp14sg5 kernel: ath11k_pci 0000:02:00.0: fw_version
> > 0x11088c35 fw_build_timestamp 2024-04-17 08:34 fw_build_id
> > WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.41
> > Mar 29 15:26:30 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-REGDOM-CHANGE init=DRIVER type=COUNTRY alpha2=US
> > Mar 29 15:26:30 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-REGDOM-CHANGE init=DRIVER type=COUNTRY alpha2=US
> > Mar 29 15:26:30 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-REGDOM-CHANGE init=DRIVER type=COUNTRY alpha2=US
> > Mar 29 15:26:35 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-CONNECTED - Connection to 68:d7:9a:2a:94:f8 completed [id=0
> > id_str=]
> > Mar 29 15:26:49 kjfp14sg5 wpa_supplicant[2373]: wlp2s0: CTRL-EVENT-BEACON-LOSS
>
> this is the reason to your disconnection
>
> > Mar 29 15:26:55 kjfp14sg5 kernel: ath11k_pci 0000:02:00.0: failed to
> > flush transmit queue, data pkts pending 9
> > Mar 29 15:26:55 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-DISCONNECTED bssid=68:d7:9a:2a:94:f8 reason=4
> > locally_generated=1
> > Mar 29 15:27:00 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-DISCONNECTED bssid=80:2a:a8:98:26:3e reason=6
> > Mar 29 15:27:05 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-DISCONNECTED bssid=74:ac:b9:df:54:36 reason=6
> > Mar 29 15:27:09 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-DISCONNECTED bssid=68:d7:9a:2a:94:f8 reason=2
> > Mar 29 15:27:09 kjfp14sg5 wpa_supplicant[2373]: wlp2s0:
> > CTRL-EVENT-SSID-TEMP-DISABLED id=0 ssid="Batman" auth_failures=1
> > duration=10 reason=CONN_FAILED
>
> and the bssid is disabled so association to this AP won't happen in a period.
>
> Anyway, although it works, using the PM quirk seems not the right fix. As you mentioned it
> seems like a regression starting to show in 6.17.4, can you do regression test to locate
> the issue commit?

It took a bit of trial and error, but I'm reasonably confident that [3] is
the culprit. The issue is pretty sporadic, but I've been able to reproduce
it twice in ~50 sleep/wake cycles on that commit, while the immediate
ancestor has been clean for 17 cycles so far. I'll continue testing and
report back.

The change looks suspicious as it modifies a code path that runs on every
sleep/wake cycle, and the failure appears to be some sort of corrupted
firmware state. In a couple of cases the WiFi firmware crashed outright
(MHI_CB_EE_RDDM) while the issue was occurring.

[3] 79266fd78df1 ("wifi: ath11k: HAL SRNG: don't deinitialize and
re-initialize again")

>
> >
> >>
> >>> were not previously included in the ath11k PM quirk table.
> >>>
> >>> Add DMI matches for product IDs 21ME and 21MF to apply the existing
> >>> ATH11K_PM_WOW override, improving suspend/resume behavior on these
> >>> systems.
> >>>
> >>> Tested on a ThinkPad P14s Gen 5 AMD (21ME) running 6.19.9.
> >>>
> >>> [1] https://bugzilla.kernel.org/show_bug.cgi?id=219196
> >>> [2] https://pcsupport.lenovo.com/us/en/products/laptops-and-netbooks/thinkpad-p-series-laptops/thinkpad-p14s-gen-5-type-21me-21mf/
> >>>
> >>> Fixes: ce8669a27016 ("wifi: ath11k: determine PM policy based on machine model")
> >>> Cc: stable at vger.kernel.org
> >>> Signed-off-by: Kyle Farnung <kfarnung at gmail.com>
> >>> ---
> >>> Changes in v2:
> >>> - Fix missing mailing list recipients (linux-wireless, ath11k, linux-kernel)
> >>> - Link to v1: https://lore.kernel.org/r/20260330-p14s-pm-quirk-v1-1-cf2fa39cc2d5@gmail.com
> >>> ---
> >>>  drivers/net/wireless/ath/ath11k/core.c | 14 ++++++++++++++
> >>>  1 file changed, 14 insertions(+)
> >>>
> >>> diff --git a/drivers/net/wireless/ath/ath11k/core.c b/drivers/net/wireless/ath/ath11k/core.c
> >>> index 3f6f4db5b7ee1aba79fd7526e5d59d068e0f4a2e..21d366224e75904feeae6cb9c93d9ef692d127fe 100644
> >>> --- a/drivers/net/wireless/ath/ath11k/core.c
> >>> +++ b/drivers/net/wireless/ath/ath11k/core.c
> >>> @@ -1041,6 +1041,20 @@ static const struct dmi_system_id ath11k_pm_quirk_table[] = {
> >>>                       DMI_MATCH(DMI_PRODUCT_NAME, "21D5"),
> >>>               },
> >>>       },
> >>> +     {
> >>> +             .driver_data = (void *)ATH11K_PM_WOW,
> >>> +             .matches = { /* P14s G5 AMD #1 */
> >>> +                     DMI_MATCH(DMI_BOARD_VENDOR, "LENOVO"),
> >>> +                     DMI_MATCH(DMI_PRODUCT_NAME, "21ME"),
> >>> +             },
> >>> +     },
> >>> +     {
> >>> +             .driver_data = (void *)ATH11K_PM_WOW,
> >>> +             .matches = { /* P14s G5 AMD #2 */
> >>> +                     DMI_MATCH(DMI_BOARD_VENDOR, "LENOVO"),
> >>> +                     DMI_MATCH(DMI_PRODUCT_NAME, "21MF"),
> >>> +             },
> >>> +     },
> >>>       {}
> >>>  };
> >>>
> >>>
> >>> ---
> >>> base-commit: dbd94b9831bc52a1efb7ff3de841ffc3457428ce
> >>> change-id: 20260330-p14s-pm-quirk-0a51ba19235f
> >>>
> >>> Best regards,
> >>
>