Re: [PATCH rtw-next v5 05/10] wifi: rtw89: use AP self role for monitor interfaces
From: Enver Ikromov
Date: Tue Nov 17 2026 - 04:42:54 EST
On Tue, Nov 17, 2026 at 00:41:28 +0000, Ping-Ke Shih wrote:
> But I can use your python script to inject packets without any change on
> kernel 7.1.5. Can you describe the "raw frames" more specific?
You are right. I ran a clean A/B test on RTL8852AE (stock 7.1.5 and
rtw-next) comparing the default monitor role (SELF_ROLE_CLIENT with
NET_TYPE_NO_LINK) against SELF_ROLE_AP across Probe Request, NullFunc,
16-byte RTS, Data, and QoS Data on 2.4 GHz and 5 GHz, both right after
interface up and 12 seconds later.
Without this patch, R_AX_MACID_SLEEP_0 (0xC2C0) and
R_AX_SS_MACID_PAUSE_0 (0x9EB0) remain 0, TX status reports zero
macid_drop, and live over-the-air ProbeResp, CTS, and ACK counts from
the AP are identical to SELF_ROLE_AP.
My earlier test was misled by two separate effects:
1. After monitor down/up on the same channel without the channel
restore fix, PHY0 stayed unconfigured so nothing went out on the
target channel.
2. On 7.1.5 (before commit a371139d8be9 set B_AX_SNIFFER_MODE), frames
injected with a spoofed source MAC were transmitted fine, but the
AP's unicast ProbeResp/CTS/ACK replies to that address were dropped
on RX (0xCE20 = 0x03174438), making TX appear silent.
So RTW89_SELF_ROLE_AP is unnecessary and I will drop patch 05/10 in v6.
Thanks,
Enver