diff mbox series

[V8,1/3] ath11k: switch to using ieee80211_tx_status_ext()

Message ID 1616105397-1482-2-git-send-email-pradeepc@codeaurora.org (mailing list archive)
State Changes Requested
Delegated to: Kalle Valo
Headers show
Series ath11k: add HE radiotap header support | expand

Commit Message

Pradeep Kumar Chitrapu March 18, 2021, 10:09 p.m. UTC
This allows us to pass HE rates down into the stack.

Co-developed-by: Miles Hu <milehu@codeaurora.org>
Signed-off-by: Miles Hu <milehu@codeaurora.org>
Signed-off-by: John Crispin <john@phrozen.org>
Signed-off-by: Pradeep Kumar Chitrapu <pradeepc@codeaurora.org>
---
 drivers/net/wireless/ath/ath11k/dp_tx.c | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)

Comments

Karthikeyan periyasamy March 19, 2021, 5:31 a.m. UTC | #1
> -	ieee80211_tx_status(ar->hw, msdu);
> +	spin_lock_bh(&ab->base_lock);
> +	peer = ath11k_peer_find_by_id(ab, ts->peer_id);
> +	if (peer) {
> +		arsta = (struct ath11k_sta *)peer->sta->drv_priv;
> +		status.sta = peer->sta;
> +		status.skb = msdu;
> +		status.info = info;
> +		status.rate = &arsta->last_txrate;

Assigning arsta holded last_txrate pointer to status.rate create race 
condition problem b/w sta delete and ieee80211_tx_status_ext, no ?
Hw we ensure that arsta pointer is valid until ieee80211_tx_status_ext() 
processing?

Instead why don't we have local struct rate_info and assign like below 
code snippet

         struct rate_info rate;
...
         rate = arsta->last_txrate;
         status.rate = &rate;

Thanks,
Karthikeyan P
Pradeep Kumar Chitrapu April 4, 2021, 12:56 p.m. UTC | #2
On 2021-03-18 22:31, Karthikeyan periyasamy wrote:
>> -	ieee80211_tx_status(ar->hw, msdu);
>> +	spin_lock_bh(&ab->base_lock);
>> +	peer = ath11k_peer_find_by_id(ab, ts->peer_id);
>> +	if (peer) {
>> +		arsta = (struct ath11k_sta *)peer->sta->drv_priv;
>> +		status.sta = peer->sta;
>> +		status.skb = msdu;
>> +		status.info = info;
>> +		status.rate = &arsta->last_txrate;
> 
> Assigning arsta holded last_txrate pointer to status.rate create race
> condition problem b/w sta delete and ieee80211_tx_status_ext, no ?
> Hw we ensure that arsta pointer is valid until
> ieee80211_tx_status_ext() processing?
> 
> Instead why don't we have local struct rate_info and assign like below
> code snippet
> 
>         struct rate_info rate;
> ...
>         rate = arsta->last_txrate;
>         status.rate = &rate;
> 
> Thanks,
> Karthikeyan P
Thanks Karthikeyan, I have addressed this in V9..
diff mbox series

Patch

diff --git a/drivers/net/wireless/ath/ath11k/dp_tx.c b/drivers/net/wireless/ath/ath11k/dp_tx.c
index 8bba5234f81f..99fc54ee6a78 100644
--- a/drivers/net/wireless/ath/ath11k/dp_tx.c
+++ b/drivers/net/wireless/ath/ath11k/dp_tx.c
@@ -417,9 +417,12 @@  static void ath11k_dp_tx_complete_msdu(struct ath11k *ar,
 				       struct sk_buff *msdu,
 				       struct hal_tx_status *ts)
 {
+	struct ieee80211_tx_status status = { 0 };
 	struct ath11k_base *ab = ar->ab;
 	struct ieee80211_tx_info *info;
 	struct ath11k_skb_cb *skb_cb;
+	struct ath11k_peer *peer;
+	struct ath11k_sta *arsta;
 
 	if (WARN_ON_ONCE(ts->buf_rel_source != HAL_WBM_REL_SRC_MODULE_TQM)) {
 		/* Must not happen */
@@ -483,13 +486,23 @@  static void ath11k_dp_tx_complete_msdu(struct ath11k *ar,
 		ath11k_dp_tx_cache_peer_stats(ar, msdu, ts);
 	}
 
-	/* NOTE: Tx rate status reporting. Tx completion status does not have
-	 * necessary information (for example nss) to build the tx rate.
-	 * Might end up reporting it out-of-band from HTT stats.
-	 */
-
-	ieee80211_tx_status(ar->hw, msdu);
+	spin_lock_bh(&ab->base_lock);
+	peer = ath11k_peer_find_by_id(ab, ts->peer_id);
+	if (peer) {
+		arsta = (struct ath11k_sta *)peer->sta->drv_priv;
+		status.sta = peer->sta;
+		status.skb = msdu;
+		status.info = info;
+		status.rate = &arsta->last_txrate;
+	}
+	spin_unlock_bh(&ab->base_lock);
+	rcu_read_unlock();
 
+	if (peer)
+		ieee80211_tx_status_ext(ar->hw, &status);
+	else
+		dev_kfree_skb_any(msdu);
+	return;
 exit:
 	rcu_read_unlock();
 }