ath9k: Fix drain txq failure in flush

While draining the txq in flush, the buffers can be
added into the tx queue by tx_tasklet which leads to
unneccesary chip reset.

This issue was originially found with AR9382 and
running heavy uplink udp traffic with higher bandwidth
and doing frequent bgscan.

Cc: stable@kernel.org
Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

diff --git a/drivers/net/wireless/ath/ath9k/main.c b/drivers/net/wireless/ath/ath9k/main.c
index c3dbf2661a3ff826815ac41962020ba6b260898a..efdafd28fab0480f3e2116322826bd5a91cba823 100644 (file)
--- a/drivers/net/wireless/ath/ath9k/main.c
+++ b/drivers/net/wireless/ath/ath9k/main.c
@@ -2261,6 +2261,7 @@ static void ath9k_flush(struct ieee80211_hw *hw, bool drop)
        struct ath_softc *sc = hw->priv;
        int timeout = 200; /* ms */
        int i, j;
+       bool drain_txq;
 
        mutex_lock(&sc->mutex);
        cancel_delayed_work_sync(&sc->tx_complete_work);
@@ -2286,7 +2287,10 @@ static void ath9k_flush(struct ieee80211_hw *hw, bool drop)
        }
 
        ath9k_ps_wakeup(sc);
-       if (!ath_drain_all_txq(sc, false))
+       spin_lock_bh(&sc->sc_pcu_lock);
+       drain_txq = ath_drain_all_txq(sc, false);
+       spin_unlock_bh(&sc->sc_pcu_lock);
+       if (!drain_txq)
                ath_reset(sc, false);
        ath9k_ps_restore(sc);
        ieee80211_wake_queues(hw);