wifi: ath12k: fix false positive RCU warning on PREEMPT_RT
Subsystems (2)
PATCH 1/1 wifi: ath12k: fix false positive RCU warning on PREEMPT_RT
ath12k_mac_get_arvif() asserts that the caller holds an RCU read lock:
WARN_ON(!rcu_read_lock_any_held());
On PREEMPT_RT kernels built with CONFIG_PROVE_RCU=n, this produces a false positive splat whenever firmware stats events are processed, even though the caller does hold the RCU read lock.
Root cause:
- On !PROVE_RCU, rcu_read_lock_any_held() is a static inline that returns !preemptible() as a proxy for "in an RCU read section".
- On PREEMPT_RT, rcu_read_lock() does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.
- ath12k_wmi_tlv_rssi_chain_parse() holds the RCU read lock via guard(rcu)() before calling ath12k_mac_get_arvif(), so the warning is incorrect.
Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at
ath12k_mac_get_arvif+0x9e/0xd0 [ath12k]
Tainted: G W O 6.19.13-rt #1 PREEMPT_RT
Call Trace:
ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k]
ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k]
ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k]
ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k]
ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k]
ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k]
ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Replace the WARN_ON() with RCU_LOCKDEP_WARN(), which is gated on debug_lockdep_rcu_enabled() and therefore compiles out entirely when PROVE_RCU is disabled. PROVE_RCU kernels continue to get the full lockdep-based check.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
--- drivers/net/wireless/ath/ath12k/mac.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/net/wireless/ath/ath12k/mac.c +2 -1
@@ -788,7 +788,8 @@ struct ath12k_link_vif *ath12k_mac_get_arvif(struct ath12k *ar, u32 vdev_id)/* To use the arvif returned, caller must have held rcu read lock. */ -WARN_ON (!rcu_read_lock_any_held ()); +RCU_LOCKDEP_WARN (!rcu_read_lock_any_held (), +"RCU read lock not held" ); arvif_iter.vdev_id = vdev_id; arvif_iter.ar = ar;
Signature · 1 line
2.53.0
Quoted 53 lines · Yu-Hsiang Tseng
ath12k_mac_get_arvif()asserts that the caller holds an RCU read lock:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever firmware stats events are processed, even though the caller does hold the RCU read lock.Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.-
ath12k_wmi_tlv_rssi_chain_parse()holds the RCU read lock via guard(rcu)() before callingath12k_mac_get_arvif(), so the warning is incorrect.Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Replace the
WARN_ON()withRCU_LOCKDEP_WARN(), which is gated ondebug_lockdep_rcu_enabled()and therefore compiles out entirely whenPROVE_RCUis disabled.PROVE_RCUkernels continue to get the full lockdep-based check.Fixes 3dd2c68f206e ("wifi: ath12k: prepare vif data structure for MLO handling")Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
Signed-off-by Yu-Hsiang Tseng <[email protected]>--- drivers/net/wireless/ath/ath12k/mac.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)drivers/net/wireless/ath/ath12k/mac.c +0
-0
drivers/net/wireless/ath/ath12k/mac.c struct ath12k_link_vif *ath12k_mac_get_arvif(struct ath12k *ar, u32 vdev_id) · line 792
Quoted 6 lines
drivers/net/wireless/ath/ath12k/mac.c +2
-1@@ -788,7 +788,8 @@ struct ath12k_link_vif *ath12k_mac_get_arvif(struct ath12k *ar, u32 vdev_id)/* To use the arvif returned, caller must have held rcu read lock. */ -WARN_ON (!rcu_read_lock_any_held ()); +RCU_LOCKDEP_WARN (!rcu_read_lock_any_held (), +"RCU read lock not held" );
should the instance in ath12k_p2p_noa_update_vdev_iter() also need to be replaced?
Quoted 3 lines
drivers/net/wireless/ath/ath12k/mac.c +0
-0arvif_iter.vdev_id = vdev_id; arvif_iter.ar = ar;
PATCH v2 1/1 wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
Two functions in ath12k assert that the caller holds an RCU read lock:ath12k_mac_get_arvif() and ath12k_p2p_noa_update_vdev_iter(). Both use:
WARN_ON(!rcu_read_lock_any_held());
On PREEMPT_RT kernels built with CONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).
Root cause:
- On !PROVE_RCU, rcu_read_lock_any_held() is a static inline that returns !preemptible() as a proxy for "in an RCU read section".
- On PREEMPT_RT, rcu_read_lock() does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.
- Callers such as ath12k_wmi_tlv_rssi_chain_parse() (via guard(rcu)()) and ieee80211_iterate_active_interfaces_atomic() do hold the RCU read lock, so these warnings are incorrect.
Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at
ath12k_mac_get_arvif+0x9e/0xd0 [ath12k]
Tainted: G W O 6.19.13-rt #1 PREEMPT_RT
Call Trace:
ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k]
ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k]
ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k]
ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k]
ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k]
ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k]
ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Replace the WARN_ON() with RCU_LOCKDEP_WARN(), which is gated on debug_lockdep_rcu_enabled() and therefore compiles out entirely when PROVE_RCU is disabled. PROVE_RCU kernels continue to get the full lockdep-based check.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
---
Changes in v2:
- Also fix the same WARN_ON() in ath12k_p2p_noa_update_vdev_iter() (suggested by Baochen Qiang)
- Update commit message to cover both call sites
Link to v1: https://lore.kernel.org/ath12k/[email protected]/
drivers/net/wireless/ath/ath12k/mac.c | 3 ++- drivers/net/wireless/ath/ath12k/p2p.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-)
drivers/net/wireless/ath/ath12k/mac.c +2 -1
@@ -788,7 +788,8 @@ struct ath12k_link_vif *ath12k_mac_get_arvif(struct ath12k *ar, u32 vdev_id)/* To use the arvif returned, caller must have held rcu read lock. */ -WARN_ON (!rcu_read_lock_any_held ()); +RCU_LOCKDEP_WARN (!rcu_read_lock_any_held (), +"RCU read lock not held" ); arvif_iter.vdev_id = vdev_id; arvif_iter.ar = ar;
drivers/net/wireless/ath/ath12k/p2p.c +2 -1
@@ -123,7 +123,8 @@ static void ath12k_p2p_noa_update_vdev_iter(void *data, u8 *mac,struct ath12k_p2p_noa_arg *arg = data;struct ath12k_link_vif *arvif; -WARN_ON (!rcu_read_lock_any_held ()); +RCU_LOCKDEP_WARN (!rcu_read_lock_any_held (), +"RCU read lock not held" ); arvif = &ahvif->deflink;if (!arvif->is_created || arvif->ar != arg->ar || arvif->vdev_id != arg->vdev_id)return ;
Signature · 1 line
2.53.0
Re: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
Quoted 43 lines · Yu-Hsiang Tseng
Two functions in ath12k assert that the caller holds an RCU read lock:
ath12k_mac_get_arvif()andath12k_p2p_noa_update_vdev_iter(). Both use:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.- Callers such as
ath12k_wmi_tlv_rssi_chain_parse()(via guard(rcu)()) andieee80211_iterate_active_interfaces_atomic()do hold the RCU read lock, so these warnings are incorrect.Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Replace the
WARN_ON()withRCU_LOCKDEP_WARN(), which is gated ondebug_lockdep_rcu_enabled()and therefore compiles out entirely whenPROVE_RCUis disabled.PROVE_RCUkernels continue to get the full lockdep-based check.
Quoted 4 lines
Fixes 3dd2c68f206e ("wifi: ath12k: prepare vif data structure for MLO handling")Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
Suggested-by Baochen Qiang <[email protected]>Signed-off-by Yu-Hsiang Tseng <[email protected]>
Re: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
In reply to Yu-Hsiang TsengQuoted 38 lines · Yu-Hsiang Tseng
Two functions in ath12k assert that the caller holds an RCU read lock:
ath12k_mac_get_arvif()andath12k_p2p_noa_update_vdev_iter(). Both use:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.- Callers such as
ath12k_wmi_tlv_rssi_chain_parse()(via guard(rcu)()) andieee80211_iterate_active_interfaces_atomic()do hold the RCU read lock, so these warnings are incorrect.Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]
Quoted 7 lines
Replace the
WARN_ON()withRCU_LOCKDEP_WARN(), which is gated ondebug_lockdep_rcu_enabled()and therefore compiles out entirely whenPROVE_RCUis disabled.PROVE_RCUkernels continue to get the full lockdep-based check.Fixes 3dd2c68f206e ("wifi: ath12k: prepare vif data structure for MLO handling")Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
Note that Tested-on: is not a official upstream tag, it is an ath-specific tag. Since it is not an official tag, it should be specified separately from the official tags:
<commit text>
<blank line>
Tested-on: ...
<blank line>
<official upstream tags>
Signed-off-by: ...
Unless there are are other review comments there is no need to send a v3 to address this -- I can make this change when I pick up the patch.
Quoted 7 lines
Suggested-by Baochen Qiang <[email protected]>Signed-off-by Yu-Hsiang Tseng <[email protected]>---
Changes in v2:
- Also fix the sameWARN_ON()inath12k_p2p_noa_update_vdev_iter()(suggested by Baochen Qiang)
- Update commit message to cover both call sites
Note that you made v2 In-reply-to the v1 patch. In kernel.org code review that is NOT preferred. In kernel.org it is preferred that each version of a patch be a separate thread.
Quoted 35 lines
Link to v1: https://lore.kernel.org/ath12k/[email protected]/
drivers/net/wireless/ath/ath12k/mac.c | 3 ++- drivers/net/wireless/ath/ath12k/p2p.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-)drivers/net/wireless/ath/ath12k/mac.c +2
-1@@ -788,7 +788,8 @@ struct ath12k_link_vif *ath12k_mac_get_arvif(struct ath12k *ar, u32 vdev_id)/* To use the arvif returned, caller must have held rcu read lock. */ -WARN_ON (!rcu_read_lock_any_held ()); +RCU_LOCKDEP_WARN (!rcu_read_lock_any_held (), +"RCU read lock not held" ); arvif_iter.vdev_id = vdev_id; arvif_iter.ar = ar;drivers/net/wireless/ath/ath12k/p2p.c +2
-1@@ -123,7 +123,8 @@ static void ath12k_p2p_noa_update_vdev_iter(void *data, u8 *mac,struct ath12k_p2p_noa_arg *arg = data;struct ath12k_link_vif *arvif; -WARN_ON (!rcu_read_lock_any_held ()); +RCU_LOCKDEP_WARN (!rcu_read_lock_any_held (), +"RCU read lock not held" ); arvif = &ahvif->deflink;if (!arvif->is_created || arvif->ar != arg->ar || arvif->vdev_id != arg->vdev_id)return ;
Re: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
Quoted 58 lines · Jeff Johnson
Quoted 38 lines · Yu-Hsiang Tseng
Two functions in ath12k assert that the caller holds an RCU read lock:
ath12k_mac_get_arvif()andath12k_p2p_noa_update_vdev_iter(). Both use:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.- Callers such as
ath12k_wmi_tlv_rssi_chain_parse()(via guard(rcu)()) andieee80211_iterate_active_interfaces_atomic()do hold the RCU read lock, so these warnings are incorrect.Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Quoted 7 lines
Replace the
WARN_ON()withRCU_LOCKDEP_WARN(), which is gated ondebug_lockdep_rcu_enabled()and therefore compiles out entirely whenPROVE_RCUis disabled.PROVE_RCUkernels continue to get the full lockdep-based check.Fixes 3dd2c68f206e ("wifi: ath12k: prepare vif data structure for MLO handling")Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
Note that Tested-on: is not a official upstream tag, it is an ath-specific tag. Since it is not an official tag, it should be specified separately from the official tags:
<commit text>
<blank line>
Tested-on: ...
<blank line>
<official upstream tags>
Signed-off-by: ...
Quoted 2 lines
Unless there are are other review comments there is no need to send a v3 to address this -- I can make this change when I pick up the patch.
Amended patch in my pending branch:
https://git.kernel.org/pub/scm/linux/kernel/git/ath/ath.git/commit/?h=pending&id=32f70ee38388bb9089b9cffa9791f5ef8d5d32cd
Re: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
Quoted 2 lines · Jeff Johnson
Amended patch in my pending branch:
https://git.kernel.org/pub/scm/linux/kernel/git/ath/ath.git/commit/?h=pending&id=32f70ee38388bb9089b9cffa9791f5ef8d5d32cd
Thanks for picking it up and fixing the Tested-on placement.
Sebastian pointed out on v2 that the fix should use lockdep_assert_in_rcu_read_lock() rather than RCU_LOCKDEP_WARN(), since the former is gated on CONFIG_PROVE_RCU (same effect) but expresses the intent more precisely. He also noted that the bug is not PREEMPT_RT specific but applies to preemptible RCU in general.
I've sent v3 addressing both points:
https://lore.kernel.org/ath12k/[email protected]/
Let me know if you'd like me to do anything further.
Thanks,
Yu-Hsiang
Jeff Johnson <[email protected]> 於 2026年4月22日週三 下午9:33寫道:
Quoted 64 lines · Jeff Johnson
Quoted 58 lines · Yu-Hsiang Tseng
Quoted 38 lines
Two functions in ath12k assert that the caller holds an RCU read lock:
ath12k_mac_get_arvif()andath12k_p2p_noa_update_vdev_iter(). Both use:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.- Callers such as
ath12k_wmi_tlv_rssi_chain_parse()(via guard(rcu)()) andieee80211_iterate_active_interfaces_atomic()do hold the RCU read lock, so these warnings are incorrect.Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Quoted 7 lines
Replace the
WARN_ON()withRCU_LOCKDEP_WARN(), which is gated ondebug_lockdep_rcu_enabled()and therefore compiles out entirely whenPROVE_RCUis disabled.PROVE_RCUkernels continue to get the full lockdep-based check.Fixes 3dd2c68f206e ("wifi: ath12k: prepare vif data structure for MLO handling")Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
Note that Tested-on: is not a official upstream tag, it is an ath-specific tag. Since it is not an official tag, it should be specified separately from the official tags:
<commit text>
<blank line>
Tested-on: ...
<blank line>
<official upstream tags>
Signed-off-by: ...Quoted 2 lines
Unless there are are other review comments there is no need to send a v3 to address this -- I can make this change when I pick up the patch.
Amended patch in my pending branch:
https://git.kernel.org/pub/scm/linux/kernel/git/ath/ath.git/commit/?h=pending&id=32f70ee38388bb9089b9cffa9791f5ef8d5d32cd
Re: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
In reply to Yu-Hsiang TsengQuoted 43 lines · Yu-Hsiang Tseng
Two functions in ath12k assert that the caller holds an RCU read lock:
ath12k_mac_get_arvif()andath12k_p2p_noa_update_vdev_iter(). Both use:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.- Callers such as
ath12k_wmi_tlv_rssi_chain_parse()(via guard(rcu)()) andieee80211_iterate_active_interfaces_atomic()do hold the RCU read lock, so these warnings are incorrect.Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]Replace the
WARN_ON()withRCU_LOCKDEP_WARN(), which is gated ondebug_lockdep_rcu_enabled()and therefore compiles out entirely whenPROVE_RCUis disabled.PROVE_RCUkernels continue to get the full lockdep-based check.
Quoted 4 lines
Fixes 3dd2c68f206e ("wifi: ath12k: prepare vif data structure for MLO handling")Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00302-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.115823.3
Suggested-by Baochen Qiang <[email protected]>Signed-off-by Yu-Hsiang Tseng <[email protected]>
Sebastian Andrzej Siewior
maintainerRe: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
In reply to Yu-Hsiang TsengQuoted 9 lines · Yu-Hsiang Tseng
Two functions in ath12k assert that the caller holds an RCU read lock:
ath12k_mac_get_arvif()andath12k_p2p_noa_update_vdev_iter(). Both use:WARN_ON(!rcu_read_lock_any_held());On
PREEMPT_RTkernels built withCONFIG_PROVE_RCU=n, this produces a false positive splat whenever these functions are invoked from paths that do hold the RCU read lock (e.g. firmware stats processing or mac80211 interface iteration).
It depends what RCU section is expected/ tested for. SMP+preempt can use preemptible RCU which does not disable preemption either. So this is not PREEMPT_RT specific. It would be PREEMPT_RT specific if the RCU section is implied by NAPI processing to so.
Quoted 8 lines
Root cause:
- On !PROVE_RCU,
rcu_read_lock_any_held()is a static inline that returns !preemptible() as a proxy for "in an RCU read section".- On
PREEMPT_RT,rcu_read_lock()does not disable preemption. A task can therefore be preemptible while legitimately holding an RCU read lock.
As elaborated above, this is not PREEMPT_RT specific but preemptible TREE RCU.
Quoted 3 lines
- Callers such as
ath12k_wmi_tlv_rssi_chain_parse()(via guard(rcu)()) andieee80211_iterate_active_interfaces_atomic()do hold the RCU read lock, so these warnings are incorrect.
If both this then use this then I guess something like
lockdep_assert_in_rcu_read_lock()
is what you look for.
Quoted 14 lines
Typical splat seen on a WCN7850 station with periodic fw stats processing:
WARNING: drivers/net/wireless/ath/ath12k/mac.c:791 at ath12k_mac_get_arvif+0x9e/0xd0 [ath12k] Tainted: G W O 6.19.13-rt #1 PREEMPT_RT Call Trace: ath12k_wmi_tlv_rssi_chain_parse+0x69/0x170 [ath12k] ath12k_wmi_tlv_iter+0x7f/0x120 [ath12k] ath12k_wmi_tlv_fw_stats_parse+0x342/0x6b0 [ath12k] ath12k_wmi_op_rx+0xe9e/0x3150 [ath12k] ath12k_htc_rx_completion_handler+0x3df/0x5b0 [ath12k] ath12k_ce_per_engine_service+0x325/0x3e0 [ath12k] ath12k_pci_ce_workqueue+0x20/0x40 [ath12k]
If that is the call chain and there are no spinlocks involved then PREEMPT+SMP+lockdep should trigger with this patch, too. The suggestion above restricts this to lockdep only but your patch does so to.
Sebastian
Re: [PATCH v2] wifi: ath12k: fix false positive RCU warnings on PREEMPT_RT
Quoted 2 lines · Sebastian Andrzej Siewior
As elaborated above, this is not
PREEMPT_RTspecific but preemptible TREE RCU.
You're right. Sent v3 that clarifies the scope (preemptible RCU in general, not only PREEMPT_RT).
Quoted 3 lines
If both this then use this then I guess something like
lockdep_assert_in_rcu_read_lock()
is what you look for.
Adopted in v3 - thanks for the pointer, it reads much cleaner than RCU_LOCKDEP_WARN().
v3: https://lore.kernel.org/ath12k/[email protected]/
Thanks,
Yu-Hsiang
Quoted 2 lines · Baochen Qiang
should the instance in
ath12k_p2p_noa_update_vdev_iter()also need to be replaced?
Good catch, thanks. Sent v2 that also converts the WARN_ON() in ath12k_p2p_noa_update_vdev_iter() to RCU_LOCKDEP_WARN().
v2: https://lore.kernel.org/ath12k/[email protected]/
Thanks,
Yu-Hsiang