Re: [PATCH rtw-next v2] wifi: rtw88: program the beacon interval
From: Luka Gejak
Date: Tue Nov 17 2026 - 07:30:27 EST
On November 17, 2026 11:14:19 AM GMT+01:00, Mehmet Fide
<mehmet.fide@xxxxxxxxx> wrote:
>From: Mehmet Fide <mehmet.fide@xxxxxxxxxxxxxxxxxx>
>
>The driver never writes the beacon interval register, so the hardware
>keeps its default of 100 TU. An AP or IBSS started with another
>interval says that interval in its beacons and sends them 100 TU apart.
>
>With beacon_int=200 in hostapd the beacons of an RTL8822CE, RTL8822CU,
>RTL8822BU and RTL8821CU are 102.4 ms apart on the air. The same holds
>for an IBSS joined with a beacon interval of 200 on an RTL8822CE and
>RTL8822BU.
>
>Write the interval when mac80211 sets it for an AP or IBSS interface.
>Both beacon from port 0: rtw_core_port_switch() moves an AP there, and
>an IBSS interface is never combined with another one.
>
>Reported-by: Luka Gejak <luka.gejak@xxxxxxxxx>
>Closes: https://lore.kernel.org/linux-wireless/1D396970-35B7-4B0B-82E9-24B523AD69C2@xxxxxxxxx/
>Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver")
>Signed-off-by: Mehmet Fide <mehmet.fide@xxxxxxxxxxxxxxxxxx>
>---
>
>Notes:
> v2:
> - write the interval for an AP or IBSS interface only (Ping-Ke)
> - Luka's Reviewed-by not carried, the condition changed
> - rebased on rtw-next fb51e39dfaea
>
> Tested on rtw-next on an RTL8822CE, RTL8822CU, RTL8822BU and RTL8821CU
> as AP with hostapd, the beacons sniffed from a second machine: with
> beacon_int 50, 100 and 200 they are 51.2, 102.4 and 204.8 ms apart on
> the air. As IBSS on the RTL8822CE and RTL8822BU, joined with a beacon
> interval of 50 and 200, the TSF of their beacons shows them 50 and 200
> TU apart. AP, association, power save, firmware crash and scan tests on
> the RTL8822BU and RTL8821CU on both bands gave no new kernel report.
>
> v1: https://lore.kernel.org/r/20261116072022.1142783-1-mehmet.fide@xxxxxxxxx/
>
> drivers/net/wireless/realtek/rtw88/mac80211.c | 5 +++++
> drivers/net/wireless/realtek/rtw88/reg.h | 1 +
> 2 files changed, 6 insertions(+)
>
>diff --git a/drivers/net/wireless/realtek/rtw88/mac80211.c b/drivers/net/wireless/realtek/rtw88/mac80211.c
>index 5e40cf06a9e3..26bc4a9913be 100644
>--- a/drivers/net/wireless/realtek/rtw88/mac80211.c
>+++ b/drivers/net/wireless/realtek/rtw88/mac80211.c
>@@ -432,6 +432,11 @@ static void rtw_ops_bss_info_changed(struct ieee80211_hw *hw,
> if (changed & BSS_CHANGED_BEACON_INT) {
> if (ieee80211_vif_type_p2p(vif) == NL80211_IFTYPE_STATION)
> coex_stat->wl_beacon_interval = conf->beacon_int;
>+ else if ((vif->type == NL80211_IFTYPE_AP ||
>+ vif->type == NL80211_IFTYPE_ADHOC) &&
>+ rtwvif->port == RTW_PORT_0)
>+ rtw_write16(rtwdev, REG_MBSSID_BCN_SPACE,
>+ conf->beacon_int);
> }
>
> if (changed & BSS_CHANGED_BEACON) {
[ ... ]
As far as I can see this is the same as v1 just expressed explicitly.
Either way I am fine with it.
Reviewed-by: Luka Gejak <luka.gejak@xxxxxxxxx>
Best regards,
Luka Gejak