From: Karthik M <[email protected]>
Transmission of management packets are done in a work queue. Sometimes
the workqueue does not finish Tx immediately, then it lead after the next
step of vdev delete finished, it start to send the management packet to
firmware and lead firmware crash.
ieee80211_set_disassoc() have logic of ieee80211_flush_queues() after
it send_deauth_disassoc() to ath12k, its purpose is make sure the
deauth was actually sent, so it need to change ath12k to match the
purpose of mac80211.
To address these issues wait for Tx management as well as Tx data packets.
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Karthik M <[email protected]>
Signed-off-by: Ramya Gnanasekar <[email protected]>
diff --git a/drivers/net/wireless/ath/ath12k/core.c b/drivers/net/wireless/ath/ath12k/core.c
index a89e66653f04..499b81cd938e 100644
--- a/drivers/net/wireless/ath/ath12k/core.c
+++ b/drivers/net/wireless/ath/ath12k/core.c
@@ -706,6 +706,7 @@ static void ath12k_core_pre_reconfigure_recovery(struct ath12k_base *ab)
idr_for_each(&ar->txmgmt_idr,
ath12k_mac_tx_mgmt_pending_free, ar);
idr_destroy(&ar->txmgmt_idr);
+ wake_up(&ar->txmgmt_empty_waitq);
}
wake_up(&ab->wmi_ab.tx_credits_wq);
diff --git a/drivers/net/wireless/ath/ath12k/core.h b/drivers/net/wireless/ath/ath12k/core.h
index dffa687ee40e..509a9d4450b4 100644
--- a/drivers/net/wireless/ath/ath12k/core.h
+++ b/drivers/net/wireless/ath/ath12k/core.h
@@ -532,6 +532,7 @@ struct ath12k {
/* protects txmgmt_idr data */
spinlock_t txmgmt_idr_lock;
atomic_t num_pending_mgmt_tx;
+ wait_queue_head_t txmgmt_empty_waitq;
/* cycle count is reported twice for each visited channel during scan.
* access protected by data_lock
diff --git a/drivers/net/wireless/ath/ath12k/mac.c b/drivers/net/wireless/ath/ath12k/mac.c
index bf7e5b6977b2..0e5cc477ba56 100644
--- a/drivers/net/wireless/ath/ath12k/mac.c
+++ b/drivers/net/wireless/ath/ath12k/mac.c
@@ -4316,6 +4316,21 @@ static int __ath12k_set_antenna(struct ath12k *ar, u32 tx_ant, u32 rx_ant)
return 0;
}
+static void ath12k_mgmt_over_wmi_tx_drop(struct ath12k *ar, struct sk_buff *skb)
+{
+ int num_mgmt;
+
+ ieee80211_free_txskb(ar->hw, skb);
+
+ num_mgmt = atomic_dec_if_positive(&ar->num_pending_mgmt_tx);
+
+ if (num_mgmt < 0)
+ WARN_ON_ONCE(1);
+
+ if (!num_mgmt)
+ wake_up(&ar->txmgmt_empty_waitq);
+}
+
int ath12k_mac_tx_mgmt_pending_free(int buf_id, void *skb, void *ctx)
{
struct sk_buff *msdu = skb;
@@ -4332,7 +4347,7 @@ int ath12k_mac_tx_mgmt_pending_free(int buf_id, void *skb, void *ctx)
info = IEEE80211_SKB_CB(msdu);
memset(&info->status, 0, sizeof(info->status));
- ieee80211_free_txskb(ar->hw, msdu);
+ ath12k_mgmt_over_wmi_tx_drop(ar, skb);
return 0;
}
@@ -4416,7 +4431,7 @@ static void ath12k_mgmt_over_wmi_tx_purge(struct ath12k *ar)
struct sk_buff *skb;
while ((skb = skb_dequeue(&ar->wmi_mgmt_tx_queue)) != NULL)
- ieee80211_free_txskb(ar->hw, skb);
+ ath12k_mgmt_over_wmi_tx_drop(ar, skb);
}
static void ath12k_mgmt_over_wmi_tx_work(struct work_struct *work)
@@ -4431,7 +4446,7 @@ static void ath12k_mgmt_over_wmi_tx_work(struct work_struct *work)
skb_cb = ATH12K_SKB_CB(skb);
if (!skb_cb->vif) {
ath12k_warn(ar->ab, "no vif found for mgmt frame\n");
- ieee80211_free_txskb(ar->hw, skb);
+ ath12k_mgmt_over_wmi_tx_drop(ar, skb);
continue;
}
@@ -4442,16 +4457,14 @@ static void ath12k_mgmt_over_wmi_tx_work(struct work_struct *work)
if (ret) {
ath12k_warn(ar->ab, "failed to tx mgmt frame, vdev_id %d :%d\n",
arvif->vdev_id, ret);
- ieee80211_free_txskb(ar->hw, skb);
- } else {
- atomic_inc(&ar->num_pending_mgmt_tx);
+ ath12k_mgmt_over_wmi_tx_drop(ar, skb);
}
} else {
ath12k_warn(ar->ab,
"dropping mgmt frame for vdev %d, is_started %d\n",
arvif->vdev_id,
arvif->is_started);
- ieee80211_free_txskb(ar->hw, skb);
+ ath12k_mgmt_over_wmi_tx_drop(ar, skb);
}
}
}
@@ -4482,6 +4495,7 @@ static int ath12k_mac_mgmt_tx(struct ath12k *ar, struct sk_buff *skb,
}
skb_queue_tail(q, skb);
+ atomic_inc(&ar->num_pending_mgmt_tx);
ieee80211_queue_work(ar->hw, &ar->wmi_mgmt_tx_work);
return 0;
@@ -5955,6 +5969,13 @@ static void ath12k_mac_op_flush(struct ieee80211_hw *hw, struct ieee80211_vif *v
ATH12K_FLUSH_TIMEOUT);
if (time_left == 0)
ath12k_warn(ar->ab, "failed to flush transmit queue %ld\n", time_left);
+
+ time_left = wait_event_timeout(ar->txmgmt_empty_waitq,
+ (atomic_read(&ar->num_pending_mgmt_tx) == 0),
+ ATH12K_FLUSH_TIMEOUT);
+ if (time_left == 0)
+ ath12k_warn(ar->ab, "failed to flush mgmt transmit queue %ld\n",
+ time_left);
}
static int
@@ -6932,6 +6953,7 @@ int ath12k_mac_register(struct ath12k_base *ab)
if (ret)
goto err_cleanup;
+ init_waitqueue_head(&ar->txmgmt_empty_waitq);
idr_init(&ar->txmgmt_idr);
spin_lock_init(&ar->txmgmt_idr_lock);
}
diff --git a/drivers/net/wireless/ath/ath12k/wmi.c b/drivers/net/wireless/ath/ath12k/wmi.c
index 3e6991120e53..1d303b217758 100644
--- a/drivers/net/wireless/ath/ath12k/wmi.c
+++ b/drivers/net/wireless/ath/ath12k/wmi.c
@@ -4637,6 +4637,7 @@ static int wmi_process_mgmt_tx_comp(struct ath12k *ar, u32 desc_id,
struct sk_buff *msdu;
struct ieee80211_tx_info *info;
struct ath12k_skb_cb *skb_cb;
+ int num_mgmt;
spin_lock_bh(&ar->txmgmt_idr_lock);
msdu = idr_find(&ar->txmgmt_idr, desc_id);
@@ -4660,10 +4661,15 @@ static int wmi_process_mgmt_tx_comp(struct ath12k *ar, u32 desc_id,
ieee80211_tx_status_irqsafe(ar->hw, msdu);
+ num_mgmt = atomic_dec_if_positive(&ar->num_pending_mgmt_tx);
+
/* WARN when we received this event without doing any mgmt tx */
- if (atomic_dec_if_positive(&ar->num_pending_mgmt_tx) < 0)
+ if (num_mgmt < 0)
WARN_ON_ONCE(1);
+ if (!num_mgmt)
+ wake_up(&ar->txmgmt_empty_waitq);
+
return 0;
}
--
2.17.1
Ramya Gnanasekar <[email protected]> writes:
> From: Karthik M <[email protected]>
>
> Transmission of management packets are done in a work queue. Sometimes
> the workqueue does not finish Tx immediately, then it lead after the next
> step of vdev delete finished, it start to send the management packet to
> firmware and lead firmware crash.
>
> ieee80211_set_disassoc() have logic of ieee80211_flush_queues() after
> it send_deauth_disassoc() to ath12k, its purpose is make sure the
> deauth was actually sent, so it need to change ath12k to match the
> purpose of mac80211.
>
> To address these issues wait for Tx management as well as Tx data packets.
>
> Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
>
> Signed-off-by: Karthik M <[email protected]>
> Signed-off-by: Ramya Gnanasekar <[email protected]>
>
> diff --git a/drivers/net/wireless/ath/ath12k/core.c b/drivers/net/wireless/ath/ath12k/core.c
> index a89e66653f04..499b81cd938e 100644
Ramya, how do you submit your patches? Your patches are missing the
"---" line which breaks my scripts. I have asked this already before but
no reply from you:
https://lore.kernel.org/all/[email protected]/
Please do not ignore my questions. If you continue to do that I will
just stop taking patches from you.
--
https://patchwork.kernel.org/project/linux-wireless/list/
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
> -----Original Message-----
> From: Kalle Valo <[email protected]>
> Sent: Monday, April 17, 2023 6:26 PM
> To: Ramya Gnanasekar (Temp) (QUIC) <[email protected]>
> Cc: [email protected]; [email protected]; Karthik M
> (QUIC) <[email protected]>
> Subject: Re: [PATCH] wifi: ath12k: add wait operation for tx management
> packets for flush from mac80211
>
> WARNING: This email originated from outside of Qualcomm. Please be wary of
> any links or attachments, and do not enable macros.
>
> Ramya Gnanasekar <[email protected]> writes:
>
> > From: Karthik M <[email protected]>
> >
> > Transmission of management packets are done in a work queue. Sometimes
> > the workqueue does not finish Tx immediately, then it lead after the
> > next step of vdev delete finished, it start to send the management
> > packet to firmware and lead firmware crash.
> >
> > ieee80211_set_disassoc() have logic of ieee80211_flush_queues() after
> > it send_deauth_disassoc() to ath12k, its purpose is make sure the
> > deauth was actually sent, so it need to change ath12k to match the
> > purpose of mac80211.
> >
> > To address these issues wait for Tx management as well as Tx data packets.
> >
> > Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-
> QCAHKSWPL_SILICONZ-1
> >
> > Signed-off-by: Karthik M <[email protected]>
> > Signed-off-by: Ramya Gnanasekar <[email protected]>
> >
> > diff --git a/drivers/net/wireless/ath/ath12k/core.c
> > b/drivers/net/wireless/ath/ath12k/core.c
> > index a89e66653f04..499b81cd938e 100644
>
> Ramya, how do you submit your patches? Your patches are missing the "---" line
> which breaks my scripts. I have asked this already before but no reply from you:
>
> https://lore.kernel.org/all/[email protected]/
>
> Please do not ignore my questions. If you continue to do that I will just stop
> taking patches from you.
>
> --
> https://patchwork.kernel.org/project/linux-wireless/list/
>
> https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatch
> es
Hi Kalle,
I use "git send-email" to send the patches and the patches are generated using "git format-patch" command.
Thanks,
Ramya
> -----Original Message-----
> From: Ramya Gnanasekar <[email protected]>
> Sent: Monday, April 17, 2023 8:01 PM
> To: Kalle Valo <[email protected]>; Ramya Gnanasekar (Temp) (QUIC)
> <[email protected]>
> Cc: [email protected]; [email protected]; Karthik M
> (QUIC) <[email protected]>
> Subject: RE: [PATCH] wifi: ath12k: add wait operation for tx management
> packets for flush from mac80211
>
> > -----Original Message-----
> > From: Kalle Valo <[email protected]>
> > Sent: Monday, April 17, 2023 6:26 PM
> > To: Ramya Gnanasekar (Temp) (QUIC) <[email protected]>
> > Cc: [email protected]; [email protected];
> > Karthik M
> > (QUIC) <[email protected]>
> > Subject: Re: [PATCH] wifi: ath12k: add wait operation for tx
> > management packets for flush from mac80211
> >
> > WARNING: This email originated from outside of Qualcomm. Please be
> > wary of any links or attachments, and do not enable macros.
> >
> > Ramya Gnanasekar <[email protected]> writes:
> >
> > > From: Karthik M <[email protected]>
> > >
> > > Transmission of management packets are done in a work queue.
> > > Sometimes the workqueue does not finish Tx immediately, then it lead
> > > after the next step of vdev delete finished, it start to send the
> > > management packet to firmware and lead firmware crash.
> > >
> > > ieee80211_set_disassoc() have logic of ieee80211_flush_queues()
> > > after it send_deauth_disassoc() to ath12k, its purpose is make sure
> > > the deauth was actually sent, so it need to change ath12k to match
> > > the purpose of mac80211.
> > >
> > > To address these issues wait for Tx management as well as Tx data packets.
> > >
> > > Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-
> > QCAHKSWPL_SILICONZ-1
> > >
> > > Signed-off-by: Karthik M <[email protected]>
> > > Signed-off-by: Ramya Gnanasekar <[email protected]>
> > >
> > > diff --git a/drivers/net/wireless/ath/ath12k/core.c
> > > b/drivers/net/wireless/ath/ath12k/core.c
> > > index a89e66653f04..499b81cd938e 100644
> >
> > Ramya, how do you submit your patches? Your patches are missing the
> > "---" line which breaks my scripts. I have asked this already before but no reply
> from you:
> >
> > https://lore.kernel.org/all/[email protected]/
> >
> > Please do not ignore my questions. If you continue to do that I will
> > just stop taking patches from you.
> >
> > --
> > https://patchwork.kernel.org/project/linux-wireless/list/
> >
> > https://wireless.wiki.kernel.org/en/developers/documentation/submittin
> > gpatch
> > es
>
> Hi Kalle,
>
> I use "git send-email" to send the patches and the patches are generated using
> "git format-patch" command.
>
> Thanks,
> Ramya
Hi Kalle,
I used "git format-patch -p1" which removed "---" while generating patch. I will correct it by using "git format-patch -1" in future patches.
Also I will resend this patch with right format.
Thanks,
Ramya