| Message ID | 20240110103909.240514-1-martin.kaistra@linutronix.de |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2] wifi: rtl8xxxu: enable channel switch support | expand |
> -----Original Message----- > From: Martin Kaistra <martin.kaistra@linutronix.de> > Sent: Wednesday, January 10, 2024 6:39 PM > To: linux-wireless@vger.kernel.org > Cc: Jes Sorensen <Jes.Sorensen@gmail.com>; Kalle Valo <kvalo@kernel.org>; Ping-Ke Shih > <pkshih@realtek.com>; Bitterblue Smith <rtl8821cerfe2@gmail.com>; Sebastian Andrzej Siewior > <bigeasy@linutronix.de> > Subject: [PATCH v2] wifi: rtl8xxxu: enable channel switch support > > The CSA countdown in the beacon frames, which are sent out by firmware, > needs to get updated by the driver. To achieve this, convert > update_beacon_work to delayed_work and schedule it with the beacon > interval in case CSA is active and the countdown is not complete. > > Signed-off-by: Martin Kaistra <martin.kaistra@linutronix.de> > --- > changes v1->v2: use delayed_work instead of usleep_range > v1: https://lore.kernel.org/linux-wireless/20240108111103.121378-1-martin.kaistra@linutronix.de/ > .../net/wireless/realtek/rtl8xxxu/rtl8xxxu.h | 2 +- > .../wireless/realtek/rtl8xxxu/rtl8xxxu_core.c | 18 ++++++++++++++---- > 2 files changed, 15 insertions(+), 5 deletions(-) > [...] > @@ -7764,7 +7772,7 @@ static int rtl8xxxu_probe(struct usb_interface *interface, > spin_lock_init(&priv->rx_urb_lock); > INIT_WORK(&priv->rx_urb_wq, rtl8xxxu_rx_urb_work); > INIT_DELAYED_WORK(&priv->ra_watchdog, rtl8xxxu_watchdog_callback); > - INIT_WORK(&priv->update_beacon_work, rtl8xxxu_update_beacon_work_callback); > + INIT_DELAYED_WORK(&priv->update_beacon_work, rtl8xxxu_update_beacon_work_callback); It seems like we missed cancel priv->update_beacon_work work at rtl8xxxu_stop(). Because that isn't introduced by this patch, you can decide to do it in this patch or not. Also, 'struct work_struct c2hcmd_work;' has similar problem. This could be a problem theoretically, but not easy to see the problem though. > skb_queue_head_init(&priv->c2hcmd_queue); > > usb_set_intfdata(interface, hw); [...]
Am 11.01.24 um 02:21 schrieb Ping-Ke Shih: > > >> -----Original Message----- >> From: Martin Kaistra <martin.kaistra@linutronix.de> >> Sent: Wednesday, January 10, 2024 6:39 PM >> To: linux-wireless@vger.kernel.org >> Cc: Jes Sorensen <Jes.Sorensen@gmail.com>; Kalle Valo <kvalo@kernel.org>; Ping-Ke Shih >> <pkshih@realtek.com>; Bitterblue Smith <rtl8821cerfe2@gmail.com>; Sebastian Andrzej Siewior >> <bigeasy@linutronix.de> >> Subject: [PATCH v2] wifi: rtl8xxxu: enable channel switch support >> >> The CSA countdown in the beacon frames, which are sent out by firmware, >> needs to get updated by the driver. To achieve this, convert >> update_beacon_work to delayed_work and schedule it with the beacon >> interval in case CSA is active and the countdown is not complete. >> >> Signed-off-by: Martin Kaistra <martin.kaistra@linutronix.de> >> --- >> changes v1->v2: use delayed_work instead of usleep_range >> v1: https://lore.kernel.org/linux-wireless/20240108111103.121378-1-martin.kaistra@linutronix.de/ >> .../net/wireless/realtek/rtl8xxxu/rtl8xxxu.h | 2 +- >> .../wireless/realtek/rtl8xxxu/rtl8xxxu_core.c | 18 ++++++++++++++---- >> 2 files changed, 15 insertions(+), 5 deletions(-) >> > > [...] > >> @@ -7764,7 +7772,7 @@ static int rtl8xxxu_probe(struct usb_interface *interface, >> spin_lock_init(&priv->rx_urb_lock); >> INIT_WORK(&priv->rx_urb_wq, rtl8xxxu_rx_urb_work); >> INIT_DELAYED_WORK(&priv->ra_watchdog, rtl8xxxu_watchdog_callback); >> - INIT_WORK(&priv->update_beacon_work, rtl8xxxu_update_beacon_work_callback); >> + INIT_DELAYED_WORK(&priv->update_beacon_work, rtl8xxxu_update_beacon_work_callback); > > It seems like we missed cancel priv->update_beacon_work work at rtl8xxxu_stop(). > Because that isn't introduced by this patch, you can decide to do it in this > patch or not. Also, 'struct work_struct c2hcmd_work;' has similar problem. Makes sense to me, I will add cancel_delayed_work_sync() for a v3 of this patch. Should I send a second patch for cancelling c2hcmd_work? > > This could be a problem theoretically, but not easy to see the problem though. > > >> skb_queue_head_init(&priv->c2hcmd_queue); >> >> usb_set_intfdata(interface, hw); > > [...] >
On Thu, 2024-01-11 at 10:23 +0100, Martin Kaistra wrote: > > Should I send a second patch for cancelling c2hcmd_work? > That will be good. Thank you in advance. :-)
diff --git a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu.h b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu.h index 803c76b3209c4..03307da67c2c3 100644 --- a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu.h +++ b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu.h @@ -1900,7 +1900,7 @@ struct rtl8xxxu_priv { struct delayed_work ra_watchdog; struct work_struct c2hcmd_work; struct sk_buff_head c2hcmd_queue; - struct work_struct update_beacon_work; + struct delayed_work update_beacon_work; struct rtl8xxxu_btcoex bt_coex; struct rtl8xxxu_ra_report ra_report; struct rtl8xxxu_cfo_tracking cfo_tracking; diff --git a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_core.c b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_core.c index 5fddbd6594a24..2cacf17c2f0d7 100644 --- a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_core.c +++ b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_core.c @@ -4605,7 +4605,7 @@ static int rtl8xxxu_set_tim(struct ieee80211_hw *hw, struct ieee80211_sta *sta, { struct rtl8xxxu_priv *priv = hw->priv; - schedule_work(&priv->update_beacon_work); + schedule_delayed_work(&priv->update_beacon_work, 0); return 0; } @@ -5108,7 +5108,7 @@ rtl8xxxu_bss_info_changed(struct ieee80211_hw *hw, struct ieee80211_vif *vif, } if (changed & BSS_CHANGED_BEACON) - schedule_work(&priv->update_beacon_work); + schedule_delayed_work(&priv->update_beacon_work, 0); error: return; @@ -5727,7 +5727,7 @@ static void rtl8xxxu_send_beacon_frame(struct ieee80211_hw *hw, static void rtl8xxxu_update_beacon_work_callback(struct work_struct *work) { struct rtl8xxxu_priv *priv = - container_of(work, struct rtl8xxxu_priv, update_beacon_work); + container_of(work, struct rtl8xxxu_priv, update_beacon_work.work); struct ieee80211_hw *hw = priv->hw; struct ieee80211_vif *vif = priv->vifs[0]; @@ -5736,6 +5736,14 @@ static void rtl8xxxu_update_beacon_work_callback(struct work_struct *work) return; } + if (vif->bss_conf.csa_active) { + if (ieee80211_beacon_cntdwn_is_complete(vif)) { + ieee80211_csa_finish(vif); + return; + } + schedule_delayed_work(&priv->update_beacon_work, + msecs_to_jiffies(vif->bss_conf.beacon_int)); + } rtl8xxxu_send_beacon_frame(hw, vif); } @@ -7764,7 +7772,7 @@ static int rtl8xxxu_probe(struct usb_interface *interface, spin_lock_init(&priv->rx_urb_lock); INIT_WORK(&priv->rx_urb_wq, rtl8xxxu_rx_urb_work); INIT_DELAYED_WORK(&priv->ra_watchdog, rtl8xxxu_watchdog_callback); - INIT_WORK(&priv->update_beacon_work, rtl8xxxu_update_beacon_work_callback); + INIT_DELAYED_WORK(&priv->update_beacon_work, rtl8xxxu_update_beacon_work_callback); skb_queue_head_init(&priv->c2hcmd_queue); usb_set_intfdata(interface, hw); @@ -7825,6 +7833,8 @@ static int rtl8xxxu_probe(struct usb_interface *interface, hw->wiphy->interface_modes |= BIT(NL80211_IFTYPE_AP); hw->queues = 4; + hw->wiphy->flags |= WIPHY_FLAG_HAS_CHANNEL_SWITCH; + if (priv->fops->supports_concurrent) { hw->wiphy->iface_combinations = rtl8xxxu_combinations; hw->wiphy->n_iface_combinations = ARRAY_SIZE(rtl8xxxu_combinations);
The CSA countdown in the beacon frames, which are sent out by firmware, needs to get updated by the driver. To achieve this, convert update_beacon_work to delayed_work and schedule it with the beacon interval in case CSA is active and the countdown is not complete. Signed-off-by: Martin Kaistra <martin.kaistra@linutronix.de> --- changes v1->v2: use delayed_work instead of usleep_range v1: https://lore.kernel.org/linux-wireless/20240108111103.121378-1-martin.kaistra@linutronix.de/ .../net/wireless/realtek/rtl8xxxu/rtl8xxxu.h | 2 +- .../wireless/realtek/rtl8xxxu/rtl8xxxu_core.c | 18 ++++++++++++++---- 2 files changed, 15 insertions(+), 5 deletions(-)