| Message ID | 20240626053543.1946-1-quic_kangyang@quicinc.com (mailing list archive) |
|---|---|
| State | Under Review |
| Delegated to: | Kalle Valo |
| Headers | show |
| Series | wifi: ath11k: use work queue to process beacon tx event | expand |
On 6/25/2024 10:35 PM, Kang Yang wrote: > Commit 3a415daa3e8b ("wifi: ath11k: add P2P IE in beacon template") > from Feb 28, 2024 (linux-next), leads to the following Smatch static > checker warning: > > drivers/net/wireless/ath/ath11k/wmi.c:1742 ath11k_wmi_p2p_go_bcn_ie() > warn: sleeping in atomic context > > The reason is that ath11k_bcn_tx_status_event() will directly call might > sleep function ath11k_wmi_cmd_send() during RCU read-side critical > sections. The call trace is like: > > ath11k_bcn_tx_status_event() > -> rcu_read_lock() > -> ath11k_mac_bcn_tx_event() > -> ath11k_mac_setup_bcn_tmpl() > …… > -> ath11k_wmi_bcn_tmpl() > -> ath11k_wmi_cmd_send() > -> rcu_read_unlock() > > Commit 886433a98425 ("ath11k: add support for BSS color change") added the > ath11k_mac_bcn_tx_event(), commit 01e782c89108 ("ath11k: fix warning > of RCU usage for ath11k_mac_get_arvif_by_vdev_id()") added the RCU lock > to avoid warning but also introduced this BUG. > > Use work queue to avoid directly calling ath11k_mac_bcn_tx_event() > during RCU critical sections. No need to worry the deletion of vif. comment nit: s/worry the deletion of vif. Because/worry about the deletion of vif because/ Kalle can apply this to pending > Because cancel_work_sync() will drop the work if it doesn't start or > block vif deletion until the running work is done. > > Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.30 > > Fixes: 3a415daa3e8b ("wifi: ath11k: add P2P IE in beacon template") > Reported-by: Dan Carpenter <dan.carpenter@linaro.org> > Closes: https://lore.kernel.org/all/2d277abd-5e7b-4da0-80e0-52bd96337f6e@moroto.mountain/ > Signed-off-by: Kang Yang <quic_kangyang@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com>
Jeff Johnson <quic_jjohnson@quicinc.com> writes: > On 6/25/2024 10:35 PM, Kang Yang wrote: > >> Commit 3a415daa3e8b ("wifi: ath11k: add P2P IE in beacon template") >> from Feb 28, 2024 (linux-next), leads to the following Smatch static >> checker warning: >> >> drivers/net/wireless/ath/ath11k/wmi.c:1742 ath11k_wmi_p2p_go_bcn_ie() >> warn: sleeping in atomic context >> >> The reason is that ath11k_bcn_tx_status_event() will directly call might >> sleep function ath11k_wmi_cmd_send() during RCU read-side critical >> sections. The call trace is like: >> >> ath11k_bcn_tx_status_event() >> -> rcu_read_lock() >> -> ath11k_mac_bcn_tx_event() >> -> ath11k_mac_setup_bcn_tmpl() >> …… >> -> ath11k_wmi_bcn_tmpl() >> -> ath11k_wmi_cmd_send() >> -> rcu_read_unlock() >> >> Commit 886433a98425 ("ath11k: add support for BSS color change") added the >> ath11k_mac_bcn_tx_event(), commit 01e782c89108 ("ath11k: fix warning >> of RCU usage for ath11k_mac_get_arvif_by_vdev_id()") added the RCU lock >> to avoid warning but also introduced this BUG. >> >> Use work queue to avoid directly calling ath11k_mac_bcn_tx_event() >> during RCU critical sections. No need to worry the deletion of vif. > > comment nit: > s/worry the deletion of vif. Because/worry about the deletion of vif because/ > > Kalle can apply this to pending Yup, fixed.
diff --git a/drivers/net/wireless/ath/ath11k/core.h b/drivers/net/wireless/ath/ath11k/core.h index df24f0e409af..7122176dd91e 100644 --- a/drivers/net/wireless/ath/ath11k/core.h +++ b/drivers/net/wireless/ath/ath11k/core.h @@ -399,6 +399,7 @@ struct ath11k_vif { u8 bssid[ETH_ALEN]; struct cfg80211_bitrate_mask bitrate_mask; struct delayed_work connection_loss_work; + struct work_struct bcn_tx_work; int num_legacy_stations; int rtscts_prot_mode; int txpower; diff --git a/drivers/net/wireless/ath/ath11k/mac.c b/drivers/net/wireless/ath/ath11k/mac.c index 779009774567..7ed606aaf1a6 100644 --- a/drivers/net/wireless/ath/ath11k/mac.c +++ b/drivers/net/wireless/ath/ath11k/mac.c @@ -6599,6 +6599,16 @@ static int ath11k_mac_vdev_delete(struct ath11k *ar, struct ath11k_vif *arvif) return ret; } +static void ath11k_mac_bcn_tx_work(struct work_struct *work) +{ + struct ath11k_vif *arvif = container_of(work, struct ath11k_vif, + bcn_tx_work); + + mutex_lock(&arvif->ar->conf_mutex); + ath11k_mac_bcn_tx_event(arvif); + mutex_unlock(&arvif->ar->conf_mutex); +} + static int ath11k_mac_op_add_interface(struct ieee80211_hw *hw, struct ieee80211_vif *vif) { @@ -6637,6 +6647,7 @@ static int ath11k_mac_op_add_interface(struct ieee80211_hw *hw, arvif->vif = vif; INIT_LIST_HEAD(&arvif->list); + INIT_WORK(&arvif->bcn_tx_work, ath11k_mac_bcn_tx_work); INIT_DELAYED_WORK(&arvif->connection_loss_work, ath11k_mac_vif_sta_connection_loss_work); @@ -6879,6 +6890,7 @@ static void ath11k_mac_op_remove_interface(struct ieee80211_hw *hw, int i; cancel_delayed_work_sync(&arvif->connection_loss_work); + cancel_work_sync(&arvif->bcn_tx_work); mutex_lock(&ar->conf_mutex); diff --git a/drivers/net/wireless/ath/ath11k/wmi.c b/drivers/net/wireless/ath/ath11k/wmi.c index 38f175dd1557..2662092ee00a 100644 --- a/drivers/net/wireless/ath/ath11k/wmi.c +++ b/drivers/net/wireless/ath/ath11k/wmi.c @@ -7404,7 +7404,9 @@ static void ath11k_bcn_tx_status_event(struct ath11k_base *ab, struct sk_buff *s rcu_read_unlock(); return; } - ath11k_mac_bcn_tx_event(arvif); + + queue_work(ab->workqueue, &arvif->bcn_tx_work); + rcu_read_unlock(); }
Commit 3a415daa3e8b ("wifi: ath11k: add P2P IE in beacon template") from Feb 28, 2024 (linux-next), leads to the following Smatch static checker warning: drivers/net/wireless/ath/ath11k/wmi.c:1742 ath11k_wmi_p2p_go_bcn_ie() warn: sleeping in atomic context The reason is that ath11k_bcn_tx_status_event() will directly call might sleep function ath11k_wmi_cmd_send() during RCU read-side critical sections. The call trace is like: ath11k_bcn_tx_status_event() -> rcu_read_lock() -> ath11k_mac_bcn_tx_event() -> ath11k_mac_setup_bcn_tmpl() …… -> ath11k_wmi_bcn_tmpl() -> ath11k_wmi_cmd_send() -> rcu_read_unlock() Commit 886433a98425 ("ath11k: add support for BSS color change") added the ath11k_mac_bcn_tx_event(), commit 01e782c89108 ("ath11k: fix warning of RCU usage for ath11k_mac_get_arvif_by_vdev_id()") added the RCU lock to avoid warning but also introduced this BUG. Use work queue to avoid directly calling ath11k_mac_bcn_tx_event() during RCU critical sections. No need to worry the deletion of vif. Because cancel_work_sync() will drop the work if it doesn't start or block vif deletion until the running work is done. Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.30 Fixes: 3a415daa3e8b ("wifi: ath11k: add P2P IE in beacon template") Reported-by: Dan Carpenter <dan.carpenter@linaro.org> Closes: https://lore.kernel.org/all/2d277abd-5e7b-4da0-80e0-52bd96337f6e@moroto.mountain/ Signed-off-by: Kang Yang <quic_kangyang@quicinc.com> --- drivers/net/wireless/ath/ath11k/core.h | 1 + drivers/net/wireless/ath/ath11k/mac.c | 12 ++++++++++++ drivers/net/wireless/ath/ath11k/wmi.c | 4 +++- 3 files changed, 16 insertions(+), 1 deletion(-) base-commit: cac9bfd02678adbcca9a7dce770609b9f7434d37