[Cerowrt-devel] [ath9k-devel] periodic hang of ath9k

[Felix Fietkau]

On 2014-07-14 06:25, Sujith Manoharan wrote:
> Dave Taht wrote:
>> cc-ing ath9k-devel for this update on http://www.bufferbloat.net/issues/442
>> 
>> this bug, which some people (usually on macs with low signal strength)
>> can get to occur fairly rapidly, but I can't, is driving me 9 kinds of
>> crazy...
> 
> Does stock OpenWrt also have this bug, or is this specific to Cerowrt ?
After receiving some useful debug output from Antonio Quartulli (who
was able to reproduce it easily), I believe that I have tracked down
this issue to some bugs in counting pending tx frames. When frames get
pushed through the U-APSD queue for PS responses and dropped there due
to retransmissions, the counter probably does not get decremented
properly.

I've come up with an untested patch that should fix this codepath
and make it easier to verify.

If you're affected by the bug, please test this patch:
---
--- a/drivers/net/wireless/ath/ath9k/ath9k.h
+++ b/drivers/net/wireless/ath/ath9k/ath9k.h
@@ -185,7 +185,8 @@ struct ath_atx_ac {
 
 struct ath_frame_info {
 	struct ath_buf *bf;
-	int framelen;
+	u16 framelen;
+	s8 txq;
 	enum ath9k_key_type keytype;
 	u8 keyix;
 	u8 rtscts_rate;
--- a/drivers/net/wireless/ath/ath9k/xmit.c
+++ b/drivers/net/wireless/ath/ath9k/xmit.c
@@ -147,15 +147,13 @@ static void ath_set_rates(struct ieee802
 static void ath_txq_skb_done(struct ath_softc *sc, struct ath_txq *txq,
 			     struct sk_buff *skb)
 {
-	int q;
-
-	q = skb_get_queue_mapping(skb);
-	if (txq == sc->tx.uapsdq)
-		txq = sc->tx.txq_map[q];
+	struct ath_frame_info *fi = get_frame_info(skb);
+	int q = fi->txq;
 
-	if (txq != sc->tx.txq_map[q])
+	if (q < 0)
 		return;
 
+	txq = sc->tx.txq_map[q];
 	if (WARN_ON(--txq->pending_frames < 0))
 		txq->pending_frames = 0;
 
@@ -1999,6 +1997,7 @@ static void setup_frame_info(struct ieee
 		an = (struct ath_node *) sta->drv_priv;
 
 	memset(fi, 0, sizeof(*fi));
+	fi->txq = -1;
 	if (hw_key)
 		fi->keyix = hw_key->hw_key_idx;
 	else if (an && ieee80211_is_data(hdr->frame_control) && an->ps_key > 0)
@@ -2150,6 +2149,7 @@ int ath_tx_start(struct ieee80211_hw *hw
 	struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
 	struct ieee80211_sta *sta = txctl->sta;
 	struct ieee80211_vif *vif = info->control.vif;
+	struct ath_frame_info *fi = get_frame_info(skb);
 	struct ath_softc *sc = hw->priv;
 	struct ath_txq *txq = txctl->txq;
 	struct ath_atx_tid *tid = NULL;
@@ -2170,11 +2170,13 @@ int ath_tx_start(struct ieee80211_hw *hw
 	q = skb_get_queue_mapping(skb);
 
 	ath_txq_lock(sc, txq);
-	if (txq == sc->tx.txq_map[q] &&
-	    ++txq->pending_frames > sc->tx.txq_max_pending[q] &&
-	    !txq->stopped) {
-		ieee80211_stop_queue(sc->hw, q);
-		txq->stopped = true;
+	if (txq == sc->tx.txq_map[q]) {
+		fi->txq = q;
+		if (++txq->pending_frames > sc->tx.txq_max_pending[q] &&
+		    !txq->stopped) {
+			ieee80211_stop_queue(sc->hw, q);
+			txq->stopped = true;
+		}
 	}
 
 	if (txctl->an && ieee80211_is_data_present(hdr->frame_control))