Re: [PATCH rtw-next v2 6/7] wifi: rtw88: 88xxa: adapt channel sensing to receiver gain

[Luka Gejak]

Hi Bitterblue and Jie Feng,

On November 10, 2026 3:36:35 PM GMT+01:00, Bitterblue Smith
<rtl8821cerfe2@xxxxxxxxx> wrote:
>On 06/11/2026 22:27, Luka Gejak wrote:
>> November 6, 2026 at 8:46 PM, "Bitterblue Smith"
><rtl8821cerfe2@xxxxxxxxx> wrote:
>> 
>> JP. Bitterblue, is the adapter on a USB 3 port, and do you see the LPS
>> failures on 5 GHz as well?
>
> It's in a USB 2 port. I am seeing the LPS failures on 5 GHz too.
> Channel 100, 80 MHz wide.

Thanks, that rules out the USB 3 idea.

My next guess is the lower bound. Under ETSI the L2H starts 37 dB below
normal mode and is then raised to at least dm_info->l2h_th_lb, which is
only 2 above a floor measured once at power on, on 2.4 GHz at 20 MHz
with the LNA off. At 80 MHz that may be close to the noise, and with H2L
7 lower EDCCA may stay busy. In LPS the firmware can switch the RF off,
and in the watchdog the ADC mask toggle only runs after rtw_leave_lps(),
so a busy EDCCA from sleep would still block the null frame.

Bitterblue, could you try debug_mask=0x8000 to see whether L2H sits at
that bound (printed as "EDCCA L2H floor") when it fails, and maybe
EDCCA_DC_BACKOFF_88XXA raised from 2 to 8?

Jie Feng, until this is figured out, keeping 88xxa on the normal thresholds
like the vendor driver does by default, or leaving this patch
out of v3, seems to be the safest option.

Best regards,
Luka Gejak