ath9k: Fix power save count imbalance on ath_radio_enable()

Upon a failure we never call ath9k_ps_restore() on ath_radio_enable(),
this will throw off the sc->ps_usecount. When the sc->ps_usecount
is > 0 we never put the chip to full sleep. This drains battery,
and will also make the chip fail upon resume with:

ath: Starting driver with initial channel: 5745 MHz
ath: timeout (100000 us) on reg 0x7000: 0xdeadbeef & 0x00000003 != 0x00000000

This would make the chip useless upon resume.

I cannot prove this can happen but in theory it is so best to
avoid this race completely and not have users complain about
a broken device after resume.

Cc: stable@kernel.org
Cc: Paul Stewart <pstew@google.com>
Cc: Amod Bodas <amod.bodas@atheros.com>
Signed-off-by: NLuis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

drivers/net/wireless/ath/ath9k/main.c

@@ -900,8 +900,7 @@ void ath_radio_enable(struct ath_softc *sc, struct ieee80211_hw *hw)
 	ath_update_txpow(sc);
 	if (ath_startrecv(sc) != 0) {
 		ath_err(common, "Unable to restart recv logic\n");
-		spin_unlock_bh(&sc->sc_pcu_lock);
-		return;
+		goto out;
 	}
 	if (sc->sc_flags & SC_OP_BEACONS)
 		ath_beacon_config(sc, NULL);	/* restart beacons */
@@ -915,6 +914,7 @@ void ath_radio_enable(struct ath_softc *sc, struct ieee80211_hw *hw)
 	ath9k_hw_set_gpio(ah, ah->led_pin, 0);
 
 	ieee80211_wake_queues(hw);
+out:
 	spin_unlock_bh(&sc->sc_pcu_lock);
 
 	ath9k_ps_restore(sc);