Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next-2.6 into for-davem

Conflicts:
	Documentation/feature-removal-schedule.txt
	drivers/net/wireless/ipw2x00/ipw2200.c

12 files changed, 496 insertions, 97 deletions

diff --git a/drivers/net/wireless/rt2x00/rt2500usb.c b/drivers/net/wireless/rt2x00/rt2500usb.c
index 6e94356265b3..93e44c7f3a74 100644
--- a/drivers/net/wireless/rt2x00/rt2500usb.c
+++ b/drivers/net/wireless/rt2x00/rt2500usb.c
@@ -1674,10 +1674,15 @@ static int rt2500usb_probe_hw_mode(struct rt2x00_dev *rt2x00dev)
 
 	/*
 	 * Initialize all hw fields.
+	 *
+	 * Don't set IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING unless we are
+	 * capable of sending the buffered frames out after the DTIM
+	 * transmission using rt2x00lib_beacondone. This will send out
+	 * multicast and broadcast traffic immediately instead of buffering it
+	 * infinitly and thus dropping it after some time.
 	 */
 	rt2x00dev->hw->flags =
 	    IEEE80211_HW_RX_INCLUDES_FCS |
-	    IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING |
 	    IEEE80211_HW_SIGNAL_DBM |
 	    IEEE80211_HW_SUPPORTS_PS |
 	    IEEE80211_HW_PS_NULLFUNC_STACK;
diff --git a/drivers/net/wireless/rt2x00/rt2800.h b/drivers/net/wireless/rt2x00/rt2800.h
index 2edc7742a7e9..eb8b6cab9925 100644
--- a/drivers/net/wireless/rt2x00/rt2800.h
+++ b/drivers/net/wireless/rt2x00/rt2800.h
@@ -1,5 +1,6 @@
 /*
-	Copyright (C) 2004 - 2009 Ivo van Doorn <IvDoorn@gmail.com>
+	Copyright (C) 2004 - 2010 Ivo van Doorn <IvDoorn@gmail.com>
+	Copyright (C) 2010 Willow Garage <http://www.willowgarage.com>
 	Copyright (C) 2009 Alban Browaeys <prahal@yahoo.com>
 	Copyright (C) 2009 Felix Fietkau <nbd@openwrt.org>
 	Copyright (C) 2009 Luis Correia <luis.f.correia@gmail.com>
@@ -710,8 +711,14 @@
 
 /*
  * TBTT_SYNC_CFG:
+ * BCN_AIFSN: Beacon AIFSN after TBTT interrupt in slots
+ * BCN_CWMIN: Beacon CWMin after TBTT interrupt in slots
  */
 #define TBTT_SYNC_CFG			0x1118
+#define TBTT_SYNC_CFG_TBTT_ADJUST	FIELD32(0x000000ff)
+#define TBTT_SYNC_CFG_BCN_EXP_WIN	FIELD32(0x0000ff00)
+#define TBTT_SYNC_CFG_BCN_AIFSN		FIELD32(0x000f0000)
+#define TBTT_SYNC_CFG_BCN_CWMIN		FIELD32(0x00f00000)
 
 /*
  * TSF_TIMER_DW0: Local lsb TSF timer, read-only
@@ -747,16 +754,21 @@
 #define INT_TIMER_EN_GP_TIMER		FIELD32(0x00000002)
 
 /*
- * CH_IDLE_STA: channel idle time
+ * CH_IDLE_STA: channel idle time (in us)
  */
 #define CH_IDLE_STA			0x1130
 
 /*
- * CH_BUSY_STA: channel busy time
+ * CH_BUSY_STA: channel busy time on primary channel (in us)
  */
 #define CH_BUSY_STA			0x1134
 
 /*
+ * CH_BUSY_STA_SEC: channel busy time on secondary channel in HT40 mode (in us)
+ */
+#define CH_BUSY_STA_SEC			0x1138
+
+/*
  * MAC_STATUS_CFG:
  * BBP_RF_BUSY: When set to 0, BBP and RF are stable.
  *	if 1 or higher one of the 2 registers is busy.
@@ -1342,6 +1354,9 @@
  * PID_TYPE: The PID latched from the PID field in the TXWI, can be used
  *           to match a frame with its tx result (even though the PID is
  *           only 4 bits wide).
+ * PID_QUEUE: Part of PID_TYPE, this is the queue index number (0-3)
+ * PID_ENTRY: Part of PID_TYPE, this is the queue entry index number (1-3)
+ *            This identification number is calculated by ((idx % 3) + 1).
  * TX_SUCCESS: Indicates tx success (1) or failure (0)
  * TX_AGGRE: Indicates if the frame was part of an aggregate (1) or not (0)
  * TX_ACK_REQUIRED: Indicates if the frame needed to get ack'ed (1) or not (0)
@@ -1353,6 +1368,8 @@
 #define TX_STA_FIFO			0x1718
 #define TX_STA_FIFO_VALID		FIELD32(0x00000001)
 #define TX_STA_FIFO_PID_TYPE		FIELD32(0x0000001e)
+#define TX_STA_FIFO_PID_QUEUE		FIELD32(0x00000006)
+#define TX_STA_FIFO_PID_ENTRY		FIELD32(0x00000018)
 #define TX_STA_FIFO_TX_SUCCESS		FIELD32(0x00000020)
 #define TX_STA_FIFO_TX_AGGRE		FIELD32(0x00000040)
 #define TX_STA_FIFO_TX_ACK_REQUIRED	FIELD32(0x00000080)
@@ -1435,6 +1452,24 @@
 
 /*
  * Security key table memory.
+ *
+ * The pairwise key table shares some memory with the beacon frame
+ * buffers 6 and 7. That basically means that when beacon 6 & 7
+ * are used we should only use the reduced pairwise key table which
+ * has a maximum of 222 entries.
+ *
+ * ---------------------------------------------
+ * |0x4000 | Pairwise Key   | Reduced Pairwise |
+ * |       | Table          | Key Table        |
+ * |       | Size: 256 * 32 | Size: 222 * 32   |
+ * |0x5BC0 |                |-------------------
+ * |       |                | Beacon 6         |
+ * |0x5DC0 |                |-------------------
+ * |       |                | Beacon 7         |
+ * |0x5FC0 |                |-------------------
+ * |0x5FFF |                |
+ * --------------------------
+ *
  * MAC_WCID_BASE: 8-bytes (use only 6 bytes) * 256 entry
  * PAIRWISE_KEY_TABLE_BASE: 32-byte * 256 entry
  * MAC_IVEIV_TABLE_BASE: 8-byte * 256-entry
@@ -1584,7 +1619,8 @@ struct mac_iveiv_entry {
  * 2. Extract memory from FCE table for BCN 4~5
  * 3. Extract memory from Pair-wise key table for BCN 6~7
  *    It occupied those memory of wcid 238~253 for BCN 6
- *    and wcid 222~237 for BCN 7
+ *    and wcid 222~237 for BCN 7 (see Security key table memory
+ *    for more info).
  *
  * IMPORTANT NOTE: Not sure why legacy driver does this,
  * but HW_BEACON_BASE7 is 0x0200 bytes below HW_BEACON_BASE6.
@@ -1963,10 +1999,17 @@ struct mac_iveiv_entry {
  * FRAG: 1 To inform TKIP engine this is a fragment.
  * MIMO_PS: The remote peer is in dynamic MIMO-PS mode
  * TX_OP: 0:HT TXOP rule , 1:PIFS TX ,2:Backoff, 3:sifs
- * BW: Channel bandwidth 20MHz or 40 MHz
+ * BW: Channel bandwidth 0:20MHz, 1:40 MHz (for legacy rates this will
+ *     duplicate the frame to both channels).
  * STBC: 1: STBC support MCS =0-7, 2,3 : RESERVED
  * AMPDU: 1: this frame is eligible for AMPDU aggregation, the hw will
- *        aggregate consecutive frames with the same RA and QoS TID.
+ *        aggregate consecutive frames with the same RA and QoS TID. If
+ *        a frame A with the same RA and QoS TID but AMPDU=0 is queued
+ *        directly after a frame B with AMPDU=1, frame A might still
+ *        get aggregated into the AMPDU started by frame B. So, setting
+ *        AMPDU to 0 does _not_ necessarily mean the frame is sent as
+ *        MPDU, it can still end up in an AMPDU if the previous frame
+ *        was tagged as AMPDU.
  */
 #define TXWI_W0_FRAG			FIELD32(0x00000001)
 #define TXWI_W0_MIMO_PS			FIELD32(0x00000002)
@@ -1993,6 +2036,10 @@ struct mac_iveiv_entry {
  *           frame was processed. If multiple frames are aggregated together
  *           (AMPDU==1) the reported tx status will always contain the packet
  *           id of the first frame. 0: Don't report tx status for this frame.
+ * PACKETID_QUEUE: Part of PACKETID, This is the queue index (0-3)
+ * PACKETID_ENTRY: Part of PACKETID, THis is the queue entry index (1-3)
+ *                 This identification number is calculated by ((idx % 3) + 1).
+ *		   The (+1) is required to prevent PACKETID to become 0.
  */
 #define TXWI_W1_ACK			FIELD32(0x00000001)
 #define TXWI_W1_NSEQ			FIELD32(0x00000002)
@@ -2000,6 +2047,8 @@ struct mac_iveiv_entry {
 #define TXWI_W1_WIRELESS_CLI_ID		FIELD32(0x0000ff00)
 #define TXWI_W1_MPDU_TOTAL_BYTE_COUNT	FIELD32(0x0fff0000)
 #define TXWI_W1_PACKETID		FIELD32(0xf0000000)
+#define TXWI_W1_PACKETID_QUEUE		FIELD32(0x30000000)
+#define TXWI_W1_PACKETID_ENTRY		FIELD32(0xc0000000)
 
 /*
  * Word2
diff --git a/drivers/net/wireless/rt2x00/rt2800lib.c b/drivers/net/wireless/rt2x00/rt2800lib.c
index 3bb67492d754..10aefc4fb0cc 100644
--- a/drivers/net/wireless/rt2x00/rt2800lib.c
+++ b/drivers/net/wireless/rt2x00/rt2800lib.c
@@ -483,7 +483,8 @@ void rt2800_write_tx_data(struct queue_entry *entry,
 			   txdesc->key_idx : 0xff);
 	rt2x00_set_field32(&word, TXWI_W1_MPDU_TOTAL_BYTE_COUNT,
 			   txdesc->length);
-	rt2x00_set_field32(&word, TXWI_W1_PACKETID, txdesc->qid + 1);
+	rt2x00_set_field32(&word, TXWI_W1_PACKETID_QUEUE, txdesc->qid);
+	rt2x00_set_field32(&word, TXWI_W1_PACKETID_ENTRY, (entry->entry_idx % 3) + 1);
 	rt2x00_desc_write(txwi, 1, word);
 
 	/*
@@ -630,15 +631,90 @@ static bool rt2800_txdone_entry_check(struct queue_entry *entry, u32 reg)
 	return true;
 }
 
+void rt2800_txdone_entry(struct queue_entry *entry, u32 status)
+{
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
+	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	struct txdone_entry_desc txdesc;
+	u32 word;
+	u16 mcs, real_mcs;
+	int aggr, ampdu;
+	__le32 *txwi;
+
+	/*
+	 * Obtain the status about this packet.
+	 */
+	txdesc.flags = 0;
+	txwi = rt2800_drv_get_txwi(entry);
+	rt2x00_desc_read(txwi, 0, &word);
+
+	mcs = rt2x00_get_field32(word, TXWI_W0_MCS);
+	ampdu = rt2x00_get_field32(word, TXWI_W0_AMPDU);
+
+	real_mcs = rt2x00_get_field32(status, TX_STA_FIFO_MCS);
+	aggr = rt2x00_get_field32(status, TX_STA_FIFO_TX_AGGRE);
+
+	/*
+	 * If a frame was meant to be sent as a single non-aggregated MPDU
+	 * but ended up in an aggregate the used tx rate doesn't correlate
+	 * with the one specified in the TXWI as the whole aggregate is sent
+	 * with the same rate.
+	 *
+	 * For example: two frames are sent to rt2x00, the first one sets
+	 * AMPDU=1 and requests MCS7 whereas the second frame sets AMDPU=0
+	 * and requests MCS15. If the hw aggregates both frames into one
+	 * AMDPU the tx status for both frames will contain MCS7 although
+	 * the frame was sent successfully.
+	 *
+	 * Hence, replace the requested rate with the real tx rate to not
+	 * confuse the rate control algortihm by providing clearly wrong
+	 * data.
+	 */
+	if (aggr == 1 && ampdu == 0 && real_mcs != mcs) {
+		skbdesc->tx_rate_idx = real_mcs;
+		mcs = real_mcs;
+	}
+
+	/*
+	 * Ralink has a retry mechanism using a global fallback
+	 * table. We setup this fallback table to try the immediate
+	 * lower rate for all rates. In the TX_STA_FIFO, the MCS field
+	 * always contains the MCS used for the last transmission, be
+	 * it successful or not.
+	 */
+	if (rt2x00_get_field32(status, TX_STA_FIFO_TX_SUCCESS)) {
+		/*
+		 * Transmission succeeded. The number of retries is
+		 * mcs - real_mcs
+		 */
+		__set_bit(TXDONE_SUCCESS, &txdesc.flags);
+		txdesc.retry = ((mcs > real_mcs) ? mcs - real_mcs : 0);
+	} else {
+		/*
+		 * Transmission failed. The number of retries is
+		 * always 7 in this case (for a total number of 8
+		 * frames sent).
+		 */
+		__set_bit(TXDONE_FAILURE, &txdesc.flags);
+		txdesc.retry = rt2x00dev->long_retry;
+	}
+
+	/*
+	 * the frame was retried at least once
+	 * -> hw used fallback rates
+	 */
+	if (txdesc.retry)
+		__set_bit(TXDONE_FALLBACK, &txdesc.flags);
+
+	rt2x00lib_txdone(entry, &txdesc);
+}
+EXPORT_SYMBOL_GPL(rt2800_txdone_entry);
+
 void rt2800_txdone(struct rt2x00_dev *rt2x00dev)
 {
 	struct data_queue *queue;
 	struct queue_entry *entry;
-	__le32 *txwi;
-	struct txdone_entry_desc txdesc;
-	u32 word;
 	u32 reg;
-	u16 mcs, real_mcs;
 	u8 pid;
 	int i;
 
@@ -660,7 +736,7 @@ void rt2800_txdone(struct rt2x00_dev *rt2x00dev)
 		 * Skip this entry when it contains an invalid
 		 * queue identication number.
 		 */
-		pid = rt2x00_get_field32(reg, TX_STA_FIFO_PID_TYPE) - 1;
+		pid = rt2x00_get_field32(reg, TX_STA_FIFO_PID_QUEUE);
 		if (pid >= QID_RX)
 			continue;
 
@@ -673,7 +749,6 @@ void rt2800_txdone(struct rt2x00_dev *rt2x00dev)
 		 * order. We first check that the queue is not empty.
 		 */
 		entry = NULL;
-		txwi = NULL;
 		while (!rt2x00queue_empty(queue)) {
 			entry = rt2x00queue_get_entry(queue, Q_INDEX_DONE);
 			if (rt2800_txdone_entry_check(entry, reg))
@@ -683,48 +758,7 @@ void rt2800_txdone(struct rt2x00_dev *rt2x00dev)
 		if (!entry || rt2x00queue_empty(queue))
 			break;
 
-
-		/*
-		 * Obtain the status about this packet.
-		 */
-		txdesc.flags = 0;
-		txwi = rt2800_drv_get_txwi(entry);
-		rt2x00_desc_read(txwi, 0, &word);
-		mcs = rt2x00_get_field32(word, TXWI_W0_MCS);
-		real_mcs = rt2x00_get_field32(reg, TX_STA_FIFO_MCS);
-
-		/*
-		 * Ralink has a retry mechanism using a global fallback
-		 * table. We setup this fallback table to try the immediate
-		 * lower rate for all rates. In the TX_STA_FIFO, the MCS field
-		 * always contains the MCS used for the last transmission, be
-		 * it successful or not.
-		 */
-		if (rt2x00_get_field32(reg, TX_STA_FIFO_TX_SUCCESS)) {
-			/*
-			 * Transmission succeeded. The number of retries is
-			 * mcs - real_mcs
-			 */
-			__set_bit(TXDONE_SUCCESS, &txdesc.flags);
-			txdesc.retry = ((mcs > real_mcs) ? mcs - real_mcs : 0);
-		} else {
-			/*
-			 * Transmission failed. The number of retries is
-			 * always 7 in this case (for a total number of 8
-			 * frames sent).
-			 */
-			__set_bit(TXDONE_FAILURE, &txdesc.flags);
-			txdesc.retry = rt2x00dev->long_retry;
-		}
-
-		/*
-		 * the frame was retried at least once
-		 * -> hw used fallback rates
-		 */
-		if (txdesc.retry)
-			__set_bit(TXDONE_FALLBACK, &txdesc.flags);
-
-		rt2x00lib_txdone(entry, &txdesc);
+		rt2800_txdone_entry(entry, reg);
 	}
 }
 EXPORT_SYMBOL_GPL(rt2800_txdone);
@@ -1031,8 +1065,12 @@ int rt2800_config_pairwise_key(struct rt2x00_dev *rt2x00dev,
 		 * 1 pairwise key is possible per AID, this means that the AID
 		 * equals our hw_key_idx. Make sure the WCID starts _after_ the
 		 * last possible shared key entry.
+		 *
+		 * Since parts of the pairwise key table might be shared with
+		 * the beacon frame buffers 6 & 7 we should only write into the
+		 * first 222 entries.
 		 */
-		if (crypto->aid > (256 - 32))
+		if (crypto->aid > (222 - 32))
 			return -ENOSPC;
 
 		key->hw_key_idx = 32 + crypto->aid;
@@ -1159,6 +1197,102 @@ void rt2800_config_intf(struct rt2x00_dev *rt2x00dev, struct rt2x00_intf *intf,
 }
 EXPORT_SYMBOL_GPL(rt2800_config_intf);
 
+static void rt2800_config_ht_opmode(struct rt2x00_dev *rt2x00dev,
+				    struct rt2x00lib_erp *erp)
+{
+	bool any_sta_nongf = !!(erp->ht_opmode &
+				IEEE80211_HT_OP_MODE_NON_GF_STA_PRSNT);
+	u8 protection = erp->ht_opmode & IEEE80211_HT_OP_MODE_PROTECTION;
+	u8 mm20_mode, mm40_mode, gf20_mode, gf40_mode;
+	u16 mm20_rate, mm40_rate, gf20_rate, gf40_rate;
+	u32 reg;
+
+	/* default protection rate for HT20: OFDM 24M */
+	mm20_rate = gf20_rate = 0x4004;
+
+	/* default protection rate for HT40: duplicate OFDM 24M */
+	mm40_rate = gf40_rate = 0x4084;
+
+	switch (protection) {
+	case IEEE80211_HT_OP_MODE_PROTECTION_NONE:
+		/*
+		 * All STAs in this BSS are HT20/40 but there might be
+		 * STAs not supporting greenfield mode.
+		 * => Disable protection for HT transmissions.
+		 */
+		mm20_mode = mm40_mode = gf20_mode = gf40_mode = 0;
+
+		break;
+	case IEEE80211_HT_OP_MODE_PROTECTION_20MHZ:
+		/*
+		 * All STAs in this BSS are HT20 or HT20/40 but there
+		 * might be STAs not supporting greenfield mode.
+		 * => Protect all HT40 transmissions.
+		 */
+		mm20_mode = gf20_mode = 0;
+		mm40_mode = gf40_mode = 2;
+
+		break;
+	case IEEE80211_HT_OP_MODE_PROTECTION_NONMEMBER:
+		/*
+		 * Nonmember protection:
+		 * According to 802.11n we _should_ protect all
+		 * HT transmissions (but we don't have to).
+		 *
+		 * But if cts_protection is enabled we _shall_ protect
+		 * all HT transmissions using a CCK rate.
+		 *
+		 * And if any station is non GF we _shall_ protect
+		 * GF transmissions.
+		 *
+		 * We decide to protect everything
+		 * -> fall through to mixed mode.
+		 */
+	case IEEE80211_HT_OP_MODE_PROTECTION_NONHT_MIXED:
+		/*
+		 * Legacy STAs are present
+		 * => Protect all HT transmissions.
+		 */
+		mm20_mode = mm40_mode = gf20_mode = gf40_mode = 2;
+
+		/*
+		 * If erp protection is needed we have to protect HT
+		 * transmissions with CCK 11M long preamble.
+		 */
+		if (erp->cts_protection) {
+			/* don't duplicate RTS/CTS in CCK mode */
+			mm20_rate = mm40_rate = 0x0003;
+			gf20_rate = gf40_rate = 0x0003;
+		}
+		break;
+	};
+
+	/* check for STAs not supporting greenfield mode */
+	if (any_sta_nongf)
+		gf20_mode = gf40_mode = 2;
+
+	/* Update HT protection config */
+	rt2800_register_read(rt2x00dev, MM20_PROT_CFG, &reg);
+	rt2x00_set_field32(&reg, MM20_PROT_CFG_PROTECT_RATE, mm20_rate);
+	rt2x00_set_field32(&reg, MM20_PROT_CFG_PROTECT_CTRL, mm20_mode);
+	rt2800_register_write(rt2x00dev, MM20_PROT_CFG, reg);
+
+	rt2800_register_read(rt2x00dev, MM40_PROT_CFG, &reg);
+	rt2x00_set_field32(&reg, MM40_PROT_CFG_PROTECT_RATE, mm40_rate);
+	rt2x00_set_field32(&reg, MM40_PROT_CFG_PROTECT_CTRL, mm40_mode);
+	rt2800_register_write(rt2x00dev, MM40_PROT_CFG, reg);
+
+	rt2800_register_read(rt2x00dev, GF20_PROT_CFG, &reg);
+	rt2x00_set_field32(&reg, GF20_PROT_CFG_PROTECT_RATE, gf20_rate);
+	rt2x00_set_field32(&reg, GF20_PROT_CFG_PROTECT_CTRL, gf20_mode);
+	rt2800_register_write(rt2x00dev, GF20_PROT_CFG, reg);
+
+	rt2800_register_read(rt2x00dev, GF40_PROT_CFG, &reg);
+	rt2x00_set_field32(&reg, GF40_PROT_CFG_PROTECT_RATE, gf40_rate);
+	rt2x00_set_field32(&reg, GF40_PROT_CFG_PROTECT_CTRL, gf40_mode);
+	rt2800_register_write(rt2x00dev, GF40_PROT_CFG, reg);
+}
+
 void rt2800_config_erp(struct rt2x00_dev *rt2x00dev, struct rt2x00lib_erp *erp,
 		       u32 changed)
 {
@@ -1203,6 +1337,9 @@ void rt2800_config_erp(struct rt2x00_dev *rt2x00dev, struct rt2x00lib_erp *erp,
 				   erp->beacon_int * 16);
 		rt2800_register_write(rt2x00dev, BCN_TIME_CFG, reg);
 	}
+
+	if (changed & BSS_CHANGED_HT)
+		rt2800_config_ht_opmode(rt2x00dev, erp);
 }
 EXPORT_SYMBOL_GPL(rt2800_config_erp);
 
@@ -1907,8 +2044,7 @@ static int rt2800_init_registers(struct rt2x00_dev *rt2x00dev)
 
 	rt2800_register_read(rt2x00dev, MM40_PROT_CFG, &reg);
 	rt2x00_set_field32(&reg, MM40_PROT_CFG_PROTECT_RATE, 0x4084);
-	rt2x00_set_field32(&reg, MM40_PROT_CFG_PROTECT_CTRL,
-			   !rt2x00_is_usb(rt2x00dev));
+	rt2x00_set_field32(&reg, MM40_PROT_CFG_PROTECT_CTRL, 0);
 	rt2x00_set_field32(&reg, MM40_PROT_CFG_PROTECT_NAV, 1);
 	rt2x00_set_field32(&reg, MM40_PROT_CFG_TX_OP_ALLOW_CCK, 1);
 	rt2x00_set_field32(&reg, MM40_PROT_CFG_TX_OP_ALLOW_OFDM, 1);
@@ -3056,11 +3192,20 @@ int rt2800_probe_hw_mode(struct rt2x00_dev *rt2x00dev)
 	 * Initialize all hw fields.
 	 */
 	rt2x00dev->hw->flags =
-	    IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING |
 	    IEEE80211_HW_SIGNAL_DBM |
 	    IEEE80211_HW_SUPPORTS_PS |
 	    IEEE80211_HW_PS_NULLFUNC_STACK |
 	    IEEE80211_HW_AMPDU_AGGREGATION;
+	/*
+	 * Don't set IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING for USB devices
+	 * unless we are capable of sending the buffered frames out after the
+	 * DTIM transmission using rt2x00lib_beacondone. This will send out
+	 * multicast and broadcast traffic immediately instead of buffering it
+	 * infinitly and thus dropping it after some time.
+	 */
+	if (!rt2x00_is_usb(rt2x00dev))
+		rt2x00dev->hw->flags |=
+			IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING;
 
 	SET_IEEE80211_DEV(rt2x00dev->hw, rt2x00dev->dev);
 	SET_IEEE80211_PERM_ADDR(rt2x00dev->hw,
@@ -3071,12 +3216,13 @@ int rt2800_probe_hw_mode(struct rt2x00_dev *rt2x00dev)
 	 * As rt2800 has a global fallback table we cannot specify
 	 * more then one tx rate per frame but since the hw will
 	 * try several rates (based on the fallback table) we should
-	 * still initialize max_rates to the maximum number of rates
+	 * initialize max_report_rates to the maximum number of rates
 	 * we are going to try. Otherwise mac80211 will truncate our
 	 * reported tx rates and the rc algortihm will end up with
 	 * incorrect data.
 	 */
-	rt2x00dev->hw->max_rates = 7;
+	rt2x00dev->hw->max_rates = 1;
+	rt2x00dev->hw->max_report_rates = 7;
 	rt2x00dev->hw->max_rate_tries = 1;
 
 	rt2x00_eeprom_read(rt2x00dev, EEPROM_ANTENNA, &eeprom);
@@ -3333,8 +3479,12 @@ int rt2800_ampdu_action(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
 	switch (action) {
 	case IEEE80211_AMPDU_RX_START:
 	case IEEE80211_AMPDU_RX_STOP:
-		/* we don't support RX aggregation yet */
-		ret = -ENOTSUPP;
+		/*
+		 * The hw itself takes care of setting up BlockAck mechanisms.
+		 * So, we only have to allow mac80211 to nagotiate a BlockAck
+		 * agreement. Once that is done, the hw will BlockAck incoming
+		 * AMPDUs without further setup.
+		 */
 		break;
 	case IEEE80211_AMPDU_TX_START:
 		ieee80211_start_tx_ba_cb_irqsafe(vif, sta->addr, tid);
diff --git a/drivers/net/wireless/rt2x00/rt2800lib.h b/drivers/net/wireless/rt2x00/rt2800lib.h
index 600c5eb25c41..81cbc92e7857 100644
--- a/drivers/net/wireless/rt2x00/rt2800lib.h
+++ b/drivers/net/wireless/rt2x00/rt2800lib.h
@@ -153,6 +153,7 @@ void rt2800_write_tx_data(struct queue_entry *entry,
 void rt2800_process_rxwi(struct queue_entry *entry, struct rxdone_entry_desc *txdesc);
 
 void rt2800_txdone(struct rt2x00_dev *rt2x00dev);
+void rt2800_txdone_entry(struct queue_entry *entry, u32 status);
 
 void rt2800_write_beacon(struct queue_entry *entry, struct txentry_desc *txdesc);
 
diff --git a/drivers/net/wireless/rt2x00/rt2800pci.c b/drivers/net/wireless/rt2x00/rt2800pci.c
index 005ee153e0cc..85a134cd62bf 100644
--- a/drivers/net/wireless/rt2x00/rt2800pci.c
+++ b/drivers/net/wireless/rt2x00/rt2800pci.c
@@ -241,6 +241,7 @@ static void rt2800pci_clear_entry(struct queue_entry *entry)
 {
 	struct queue_entry_priv_pci *entry_priv = entry->priv_data;
 	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
 	u32 word;
 
 	if (entry->queue->qid == QID_RX) {
@@ -251,6 +252,13 @@ static void rt2800pci_clear_entry(struct queue_entry *entry)
 		rt2x00_desc_read(entry_priv->desc, 1, &word);
 		rt2x00_set_field32(&word, RXD_W1_DMA_DONE, 0);
 		rt2x00_desc_write(entry_priv->desc, 1, word);
+
+		/*
+		 * Set RX IDX in register to inform hardware that we have
+		 * handled this entry and it is available for reuse again.
+		 */
+		rt2800_register_write(rt2x00dev, RX_CRX_IDX,
+				      entry->entry_idx);
 	} else {
 		rt2x00_desc_read(entry_priv->desc, 1, &word);
 		rt2x00_set_field32(&word, TXD_W1_DMA_DONE, 1);
@@ -599,7 +607,6 @@ static void rt2800pci_kill_tx_queue(struct data_queue *queue)
 static void rt2800pci_fill_rxdone(struct queue_entry *entry,
 				  struct rxdone_entry_desc *rxdesc)
 {
-	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
 	struct queue_entry_priv_pci *entry_priv = entry->priv_data;
 	__le32 *rxd = entry_priv->desc;
 	u32 word;
@@ -641,12 +648,6 @@ static void rt2800pci_fill_rxdone(struct queue_entry *entry,
 	 * Process the RXWI structure that is at the start of the buffer.
 	 */
 	rt2800_process_rxwi(entry, rxdesc);
-
-	/*
-	 * Set RX IDX in register to inform hardware that we have handled
-	 * this entry and it is available for reuse again.
-	 */
-	rt2800_register_write(rt2x00dev, RX_CRX_IDX, entry->entry_idx);
 }
 
 /*
@@ -660,6 +661,63 @@ static void rt2800pci_wakeup(struct rt2x00_dev *rt2x00dev)
 	rt2800_config(rt2x00dev, &libconf, IEEE80211_CONF_CHANGE_PS);
 }
 
+static void rt2800pci_txdone(struct rt2x00_dev *rt2x00dev)
+{
+	struct data_queue *queue;
+	struct queue_entry *entry;
+	u32 status;
+	u8 qid;
+
+	while (!kfifo_is_empty(&rt2x00dev->txstatus_fifo)) {
+		/* Now remove the tx status from the FIFO */
+		if (kfifo_out(&rt2x00dev->txstatus_fifo, &status,
+			      sizeof(status)) != sizeof(status)) {
+			WARN_ON(1);
+			break;
+		}
+
+		qid = rt2x00_get_field32(status, TX_STA_FIFO_PID_TYPE) - 1;
+		if (qid >= QID_RX) {
+			/*
+			 * Unknown queue, this shouldn't happen. Just drop
+			 * this tx status.
+			 */
+			WARNING(rt2x00dev, "Got TX status report with "
+					   "unexpected pid %u, dropping", qid);
+			break;
+		}
+
+		queue = rt2x00queue_get_queue(rt2x00dev, qid);
+		if (unlikely(queue == NULL)) {
+			/*
+			 * The queue is NULL, this shouldn't happen. Stop
+			 * processing here and drop the tx status
+			 */
+			WARNING(rt2x00dev, "Got TX status for an unavailable "
+					   "queue %u, dropping", qid);
+			break;
+		}
+
+		if (rt2x00queue_empty(queue)) {
+			/*
+			 * The queue is empty. Stop processing here
+			 * and drop the tx status.
+			 */
+			WARNING(rt2x00dev, "Got TX status for an empty "
+					   "queue %u, dropping", qid);
+			break;
+		}
+
+		entry = rt2x00queue_get_entry(queue, Q_INDEX_DONE);
+		rt2800_txdone_entry(entry, status);
+	}
+}
+
+static void rt2800pci_txstatus_tasklet(unsigned long data)
+{
+	rt2800pci_txdone((struct rt2x00_dev *)data);
+}
+
 static irqreturn_t rt2800pci_interrupt_thread(int irq, void *dev_instance)
 {
 	struct rt2x00_dev *rt2x00dev = dev_instance;
@@ -684,13 +742,7 @@ static irqreturn_t rt2800pci_interrupt_thread(int irq, void *dev_instance)
 		rt2x00pci_rxdone(rt2x00dev);
 
 	/*
-	 * 4 - Tx done interrupt.
-	 */
-	if (rt2x00_get_field32(reg, INT_SOURCE_CSR_TX_FIFO_STATUS))
-		rt2800_txdone(rt2x00dev);
-
-	/*
-	 * 5 - Auto wakeup interrupt.
+	 * 4 - Auto wakeup interrupt.
 	 */
 	if (rt2x00_get_field32(reg, INT_SOURCE_CSR_AUTO_WAKEUP))
 		rt2800pci_wakeup(rt2x00dev);
@@ -702,10 +754,58 @@ static irqreturn_t rt2800pci_interrupt_thread(int irq, void *dev_instance)
 	return IRQ_HANDLED;
 }
 
+static void rt2800pci_txstatus_interrupt(struct rt2x00_dev *rt2x00dev)
+{
+	u32 status;
+	int i;
+
+	/*
+	 * The TX_FIFO_STATUS interrupt needs special care. We should
+	 * read TX_STA_FIFO but we should do it immediately as otherwise
+	 * the register can overflow and we would lose status reports.
+	 *
+	 * Hence, read the TX_STA_FIFO register and copy all tx status
+	 * reports into a kernel FIFO which is handled in the txstatus
+	 * tasklet. We use a tasklet to process the tx status reports
+	 * because we can schedule the tasklet multiple times (when the
+	 * interrupt fires again during tx status processing).
+	 *
+	 * Furthermore we don't disable the TX_FIFO_STATUS
+	 * interrupt here but leave it enabled so that the TX_STA_FIFO
+	 * can also be read while the interrupt thread gets executed.
+	 *
+	 * Since we have only one producer and one consumer we don't
+	 * need to lock the kfifo.
+	 */
+	for (i = 0; i < TX_ENTRIES; i++) {
+		rt2800_register_read(rt2x00dev, TX_STA_FIFO, &status);
+
+		if (!rt2x00_get_field32(status, TX_STA_FIFO_VALID))
+			break;
+
+		if (kfifo_is_full(&rt2x00dev->txstatus_fifo)) {
+			WARNING(rt2x00dev, "TX status FIFO overrun,"
+				" drop tx status report.\n");
+			break;
+		}
+
+		if (kfifo_in(&rt2x00dev->txstatus_fifo, &status,
+			     sizeof(status)) != sizeof(status)) {
+			WARNING(rt2x00dev, "TX status FIFO overrun,"
+				"drop tx status report.\n");
+			break;
+		}
+	}
+
+	/* Schedule the tasklet for processing the tx status. */
+	tasklet_schedule(&rt2x00dev->txstatus_tasklet);
+}
+
 static irqreturn_t rt2800pci_interrupt(int irq, void *dev_instance)
 {
 	struct rt2x00_dev *rt2x00dev = dev_instance;
 	u32 reg;
+	irqreturn_t ret = IRQ_HANDLED;
 
 	/* Read status and ACK all interrupts */
 	rt2800_register_read(rt2x00dev, INT_SOURCE_CSR, &reg);
@@ -717,15 +817,38 @@ static irqreturn_t rt2800pci_interrupt(int irq, void *dev_instance)
 	if (!test_bit(DEVICE_STATE_ENABLED_RADIO, &rt2x00dev->flags))
 		return IRQ_HANDLED;
 
-	/* Store irqvalue for use in the interrupt thread. */
-	rt2x00dev->irqvalue[0] = reg;
+	if (rt2x00_get_field32(reg, INT_SOURCE_CSR_TX_FIFO_STATUS))
+		rt2800pci_txstatus_interrupt(rt2x00dev);
 
-	/* Disable interrupts, will be enabled again in the interrupt thread. */
-	rt2x00dev->ops->lib->set_device_state(rt2x00dev,
-					      STATE_RADIO_IRQ_OFF_ISR);
+	if (rt2x00_get_field32(reg, INT_SOURCE_CSR_PRE_TBTT) ||
+	    rt2x00_get_field32(reg, INT_SOURCE_CSR_TBTT) ||
+	    rt2x00_get_field32(reg, INT_SOURCE_CSR_RX_DONE) ||
+	    rt2x00_get_field32(reg, INT_SOURCE_CSR_AUTO_WAKEUP)) {
+		/*
+		 * All other interrupts are handled in the interrupt thread.
+		 * Store irqvalue for use in the interrupt thread.
+		 */
+		rt2x00dev->irqvalue[0] = reg;
+
+		/*
+		 * Disable interrupts, will be enabled again in the
+		 * interrupt thread.
+		*/
+		rt2x00dev->ops->lib->set_device_state(rt2x00dev,
+						      STATE_RADIO_IRQ_OFF_ISR);
+
+		/*
+		 * Leave the TX_FIFO_STATUS interrupt enabled to not lose any
+		 * tx status reports.
+		 */
+		rt2800_register_read(rt2x00dev, INT_MASK_CSR, &reg);
+		rt2x00_set_field32(&reg, INT_MASK_CSR_TX_FIFO_STATUS, 1);
+		rt2800_register_write(rt2x00dev, INT_MASK_CSR, reg);
 
+		ret = IRQ_WAKE_THREAD;
+	}
 
-	return IRQ_WAKE_THREAD;
+	return ret;
 }
 
 /*
@@ -788,6 +911,7 @@ static int rt2800pci_probe_hw(struct rt2x00_dev *rt2x00dev)
 		__set_bit(DRIVER_REQUIRE_FIRMWARE, &rt2x00dev->flags);
 	__set_bit(DRIVER_REQUIRE_DMA, &rt2x00dev->flags);
 	__set_bit(DRIVER_REQUIRE_L2PAD, &rt2x00dev->flags);
+	__set_bit(DRIVER_REQUIRE_TXSTATUS_FIFO, &rt2x00dev->flags);
 	if (!modparam_nohwcrypt)
 		__set_bit(CONFIG_SUPPORT_HW_CRYPTO, &rt2x00dev->flags);
 	__set_bit(DRIVER_SUPPORT_LINK_TUNING, &rt2x00dev->flags);
@@ -837,6 +961,7 @@ static const struct rt2800_ops rt2800pci_rt2800_ops = {
 static const struct rt2x00lib_ops rt2800pci_rt2x00_ops = {
 	.irq_handler		= rt2800pci_interrupt,
 	.irq_handler_thread	= rt2800pci_interrupt_thread,
+	.txstatus_tasklet       = rt2800pci_txstatus_tasklet,
 	.probe_hw		= rt2800pci_probe_hw,
 	.get_firmware_name	= rt2800pci_get_firmware_name,
 	.check_firmware		= rt2800_check_firmware,
diff --git a/drivers/net/wireless/rt2x00/rt2x00.h b/drivers/net/wireless/rt2x00/rt2x00.h
index 7832a5996a8c..75ac6624bf9e 100644
--- a/drivers/net/wireless/rt2x00/rt2x00.h
+++ b/drivers/net/wireless/rt2x00/rt2x00.h
@@ -36,6 +36,7 @@
 #include <linux/mutex.h>
 #include <linux/etherdevice.h>
 #include <linux/input-polldev.h>
+#include <linux/kfifo.h>
 
 #include <net/mac80211.h>
 
@@ -457,6 +458,7 @@ struct rt2x00lib_erp {
 	short eifs;
 
 	u16 beacon_int;
+	u16 ht_opmode;
 };
 
 /*
@@ -522,6 +524,11 @@ struct rt2x00lib_ops {
 	irq_handler_t irq_handler_thread;
 
 	/*
+	 * TX status tasklet handler.
+	 */
+	void (*txstatus_tasklet) (unsigned long data);
+
+	/*
 	 * Device init handlers.
 	 */
 	int (*probe_hw) (struct rt2x00_dev *rt2x00dev);
@@ -651,6 +658,7 @@ enum rt2x00_flags {
 	DRIVER_REQUIRE_DMA,
 	DRIVER_REQUIRE_COPY_IV,
 	DRIVER_REQUIRE_L2PAD,
+	DRIVER_REQUIRE_TXSTATUS_FIFO,
 
 	/*
 	 * Driver features
@@ -884,6 +892,16 @@ struct rt2x00_dev {
 	 * and interrupt thread routine.
 	 */
 	u32 irqvalue[2];
+
+	/*
+	 * FIFO for storing tx status reports between isr and tasklet.
+	 */
+	struct kfifo txstatus_fifo;
+
+	/*
+	 * Tasklet for processing tx status reports (rt2800pci).
+	 */
+	struct tasklet_struct txstatus_tasklet;
 };
 
 /*
diff --git a/drivers/net/wireless/rt2x00/rt2x00config.c b/drivers/net/wireless/rt2x00/rt2x00config.c
index 4c7ff765a8bf..54ffb5aeb34e 100644
--- a/drivers/net/wireless/rt2x00/rt2x00config.c
+++ b/drivers/net/wireless/rt2x00/rt2x00config.c
@@ -103,6 +103,9 @@ void rt2x00lib_config_erp(struct rt2x00_dev *rt2x00dev,
 	/* Update global beacon interval time, this is needed for PS support */
 	rt2x00dev->beacon_int = bss_conf->beacon_int;
 
+	if (changed & BSS_CHANGED_HT)
+		erp.ht_opmode = bss_conf->ht_operation_mode;
+
 	rt2x00dev->ops->lib->config_erp(rt2x00dev, &erp, changed);
 }
 
diff --git a/drivers/net/wireless/rt2x00/rt2x00dev.c b/drivers/net/wireless/rt2x00/rt2x00dev.c
index 053fdd3bd720..6f442b02b83e 100644
--- a/drivers/net/wireless/rt2x00/rt2x00dev.c
+++ b/drivers/net/wireless/rt2x00/rt2x00dev.c
@@ -813,6 +813,30 @@ static int rt2x00lib_probe_hw(struct rt2x00_dev *rt2x00dev)
 		rt2x00dev->hw->extra_tx_headroom += RT2X00_ALIGN_SIZE;
 
 	/*
+	 * Allocate tx status FIFO for driver use.
+	 */
+	if (test_bit(DRIVER_REQUIRE_TXSTATUS_FIFO, &rt2x00dev->flags) &&
+	    rt2x00dev->ops->lib->txstatus_tasklet) {
+		/*
+		 * Allocate txstatus fifo and tasklet, we use a size of 512
+		 * for the kfifo which is big enough to store 512/4=128 tx
+		 * status reports. In the worst case (tx status for all tx
+		 * queues gets reported before we've got a chance to handle
+		 * them) 24*4=384 tx status reports need to be cached.
+		 */
+		status = kfifo_alloc(&rt2x00dev->txstatus_fifo, 512,
+				     GFP_KERNEL);
+		if (status)
+			return status;
+
+		/* tasklet for processing the tx status reports. */
+		tasklet_init(&rt2x00dev->txstatus_tasklet,
+			     rt2x00dev->ops->lib->txstatus_tasklet,
+			     (unsigned long)rt2x00dev);
+
+	}
+
+	/*
 	 * Register HW.
 	 */
 	status = ieee80211_register_hw(rt2x00dev->hw);
@@ -909,10 +933,8 @@ int rt2x00lib_start(struct rt2x00_dev *rt2x00dev)
 
 	/* Enable the radio */
 	retval = rt2x00lib_enable_radio(rt2x00dev);
-	if (retval) {
-		rt2x00queue_uninitialize(rt2x00dev);
+	if (retval)
 		return retval;
-	}
 
 	set_bit(DEVICE_STATE_STARTED, &rt2x00dev->flags);
 
@@ -1028,6 +1050,16 @@ void rt2x00lib_remove_dev(struct rt2x00_dev *rt2x00dev)
 	cancel_work_sync(&rt2x00dev->txdone_work);
 
 	/*
+	 * Free the tx status fifo.
+	 */
+	kfifo_free(&rt2x00dev->txstatus_fifo);
+
+	/*
+	 * Kill the tx status tasklet.
+	 */
+	tasklet_kill(&rt2x00dev->txstatus_tasklet);
+
+	/*
 	 * Uninitialize device.
 	 */
 	rt2x00lib_uninitialize(rt2x00dev);
diff --git a/drivers/net/wireless/rt2x00/rt2x00ht.c b/drivers/net/wireless/rt2x00/rt2x00ht.c
index ad3c7ff4837b..c637bcaec5f8 100644
--- a/drivers/net/wireless/rt2x00/rt2x00ht.c
+++ b/drivers/net/wireless/rt2x00/rt2x00ht.c
@@ -60,9 +60,10 @@ void rt2x00ht_create_tx_descriptor(struct queue_entry *entry,
 		 * when using more then one tx stream (>MCS7).
 		 */
 		if (tx_info->control.sta && txdesc->mcs > 7 &&
-		    (tx_info->control.sta->ht_cap.cap &
-		     (WLAN_HT_CAP_SM_PS_DYNAMIC <<
-		      IEEE80211_HT_CAP_SM_PS_SHIFT)))
+		    ((tx_info->control.sta->ht_cap.cap &
+		      IEEE80211_HT_CAP_SM_PS) >>
+		     IEEE80211_HT_CAP_SM_PS_SHIFT) ==
+		    WLAN_HT_CAP_SM_PS_DYNAMIC)
 			__set_bit(ENTRY_TXD_HT_MIMO_PS, &txdesc->flags);
 	} else {
 		txdesc->mcs = rt2x00_get_rate_mcs(hwrate->mcs);
@@ -72,9 +73,11 @@ void rt2x00ht_create_tx_descriptor(struct queue_entry *entry,
 
 
 	/*
-	 * Convert flags
+	 * This frame is eligible for an AMPDU, however, don't aggregate
+	 * frames that are intended to probe a specific tx rate.
 	 */
-	if (tx_info->flags & IEEE80211_TX_CTL_AMPDU)
+	if (tx_info->flags & IEEE80211_TX_CTL_AMPDU &&
+	    !(tx_info->flags & IEEE80211_TX_CTL_RATE_CTRL_PROBE))
 		__set_bit(ENTRY_TXD_HT_AMPDU, &txdesc->flags);
 
 	/*
@@ -84,7 +87,13 @@ void rt2x00ht_create_tx_descriptor(struct queue_entry *entry,
 		txdesc->rate_mode = RATE_MODE_HT_MIX;
 	if (txrate->flags & IEEE80211_TX_RC_GREEN_FIELD)
 		txdesc->rate_mode = RATE_MODE_HT_GREENFIELD;
-	if (txrate->flags & IEEE80211_TX_RC_40_MHZ_WIDTH)
+
+	/*
+	 * Set 40Mhz mode if necessary (for legacy rates this will
+	 * duplicate the frame to both channels).
+	 */
+	if (txrate->flags & IEEE80211_TX_RC_40_MHZ_WIDTH ||
+	    txrate->flags & IEEE80211_TX_RC_DUP_DATA)
 		__set_bit(ENTRY_TXD_HT_BW_40, &txdesc->flags);
 	if (txrate->flags & IEEE80211_TX_RC_SHORT_GI)
 		__set_bit(ENTRY_TXD_HT_SHORT_GI, &txdesc->flags);
diff --git a/drivers/net/wireless/rt2x00/rt2x00mac.c b/drivers/net/wireless/rt2x00/rt2x00mac.c
index 7862a840984a..c3c206a97d54 100644
--- a/drivers/net/wireless/rt2x00/rt2x00mac.c
+++ b/drivers/net/wireless/rt2x00/rt2x00mac.c
@@ -671,7 +671,7 @@ void rt2x00mac_bss_info_changed(struct ieee80211_hw *hw,
 	 */
 	if (changes & (BSS_CHANGED_ERP_CTS_PROT | BSS_CHANGED_ERP_PREAMBLE |
 		       BSS_CHANGED_ERP_SLOT | BSS_CHANGED_BASIC_RATES |
-		       BSS_CHANGED_BEACON_INT))
+		       BSS_CHANGED_BEACON_INT | BSS_CHANGED_HT))
 		rt2x00lib_config_erp(rt2x00dev, intf, bss_conf, changes);
 }
 EXPORT_SYMBOL_GPL(rt2x00mac_bss_info_changed);
diff --git a/drivers/net/wireless/rt2x00/rt61pci.c b/drivers/net/wireless/rt2x00/rt61pci.c
index 97b3935f615b..af548c87f108 100644
--- a/drivers/net/wireless/rt2x00/rt61pci.c
+++ b/drivers/net/wireless/rt2x00/rt61pci.c
@@ -2630,12 +2630,13 @@ static int rt61pci_probe_hw_mode(struct rt2x00_dev *rt2x00dev)
 	 * As rt61 has a global fallback table we cannot specify
 	 * more then one tx rate per frame but since the hw will
 	 * try several rates (based on the fallback table) we should
-	 * still initialize max_rates to the maximum number of rates
+	 * initialize max_report_rates to the maximum number of rates
 	 * we are going to try. Otherwise mac80211 will truncate our
 	 * reported tx rates and the rc algortihm will end up with
 	 * incorrect data.
 	 */
-	rt2x00dev->hw->max_rates = 7;
+	rt2x00dev->hw->max_rates = 1;
+	rt2x00dev->hw->max_report_rates = 7;
 	rt2x00dev->hw->max_rate_tries = 1;
 
 	/*
diff --git a/drivers/net/wireless/rt2x00/rt73usb.c b/drivers/net/wireless/rt2x00/rt73usb.c
index e22f01c1818e..9be8089317e4 100644
--- a/drivers/net/wireless/rt2x00/rt73usb.c
+++ b/drivers/net/wireless/rt2x00/rt73usb.c
@@ -2063,9 +2063,14 @@ static int rt73usb_probe_hw_mode(struct rt2x00_dev *rt2x00dev)
 
 	/*
 	 * Initialize all hw fields.
+	 *
+	 * Don't set IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING unless we are
+	 * capable of sending the buffered frames out after the DTIM
+	 * transmission using rt2x00lib_beacondone. This will send out
+	 * multicast and broadcast traffic immediately instead of buffering it
+	 * infinitly and thus dropping it after some time.
 	 */
 	rt2x00dev->hw->flags =
-	    IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING |
 	    IEEE80211_HW_SIGNAL_DBM |
 	    IEEE80211_HW_SUPPORTS_PS |
 	    IEEE80211_HW_PS_NULLFUNC_STACK;
@@ -2365,6 +2370,7 @@ static struct usb_device_id rt73usb_device_table[] = {
 	{ USB_DEVICE(0x0411, 0x00f4), USB_DEVICE_DATA(&rt73usb_ops) },
 	{ USB_DEVICE(0x0411, 0x0116), USB_DEVICE_DATA(&rt73usb_ops) },
 	{ USB_DEVICE(0x0411, 0x0119), USB_DEVICE_DATA(&rt73usb_ops) },
+	{ USB_DEVICE(0x0411, 0x0137), USB_DEVICE_DATA(&rt73usb_ops) },
 	/* CEIVA */
 	{ USB_DEVICE(0x178d, 0x02be), USB_DEVICE_DATA(&rt73usb_ops) },
 	/* CNet */