drivers/net/wireless/ath/ath10k/htt_rx.c

/*
 * Copyright (c) 2005-2011 Atheros Communications Inc.
 * Copyright (c) 2011-2013 Qualcomm Atheros, Inc.
 *
 * Permission to use, copy, modify, and/or distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */

#include "core.h"
#include "htc.h"
#include "htt.h"
#include "txrx.h"
#include "debug.h"
#include "trace.h"
#include "mac.h"

#include <linux/log2.h>

/* slightly larger than one large A-MPDU */
#define HTT_RX_RING_SIZE_MIN 128

/* roughly 20 ms @ 1 Gbps of 1500B MSDUs */
#define HTT_RX_RING_SIZE_MAX 2048

#define HTT_RX_AVG_FRM_BYTES 1000

/* ms, very conservative */
#define HTT_RX_HOST_LATENCY_MAX_MS 20

/* ms, conservative */
#define HTT_RX_HOST_LATENCY_WORST_LIKELY_MS 10

/* when under memory pressure rx ring refill may fail and needs a retry */
#define HTT_RX_RING_REFILL_RETRY_MS 50

static int ath10k_htt_rx_get_csum_state(struct sk_buff *skb);
static void ath10k_htt_txrx_compl_task(unsigned long ptr);

static int ath10k_htt_rx_ring_size(struct ath10k_htt *htt)
{
	int size;

	/*
	 * It is expected that the host CPU will typically be able to
	 * service the rx indication from one A-MPDU before the rx
	 * indication from the subsequent A-MPDU happens, roughly 1-2 ms
	 * later. However, the rx ring should be sized very conservatively,
	 * to accomodate the worst reasonable delay before the host CPU
	 * services a rx indication interrupt.
	 *
	 * The rx ring need not be kept full of empty buffers. In theory,
	 * the htt host SW can dynamically track the low-water mark in the
	 * rx ring, and dynamically adjust the level to which the rx ring
	 * is filled with empty buffers, to dynamically meet the desired
	 * low-water mark.
	 *
	 * In contrast, it's difficult to resize the rx ring itself, once
	 * it's in use. Thus, the ring itself should be sized very
	 * conservatively, while the degree to which the ring is filled
	 * with empty buffers should be sized moderately conservatively.
	 */

	/* 1e6 bps/mbps / 1e3 ms per sec = 1000 */
	size =
	    htt->max_throughput_mbps +
	    1000  /
	    (8 * HTT_RX_AVG_FRM_BYTES) * HTT_RX_HOST_LATENCY_MAX_MS;

	if (size < HTT_RX_RING_SIZE_MIN)
		size = HTT_RX_RING_SIZE_MIN;

	if (size > HTT_RX_RING_SIZE_MAX)
		size = HTT_RX_RING_SIZE_MAX;

	size = roundup_pow_of_two(size);

	return size;
}

static int ath10k_htt_rx_ring_fill_level(struct ath10k_htt *htt)
{
	int size;

	/* 1e6 bps/mbps / 1e3 ms per sec = 1000 */
	size =
	    htt->max_throughput_mbps *
	    1000  /
	    (8 * HTT_RX_AVG_FRM_BYTES) * HTT_RX_HOST_LATENCY_WORST_LIKELY_MS;

	/*
	 * Make sure the fill level is at least 1 less than the ring size.
	 * Leaving 1 element empty allows the SW to easily distinguish
	 * between a full ring vs. an empty ring.
	 */
	if (size >= htt->rx_ring.size)
		size = htt->rx_ring.size - 1;

	return size;
}

static void ath10k_htt_rx_ring_free(struct ath10k_htt *htt)
{
	struct sk_buff *skb;
	struct ath10k_skb_cb *cb;
	int i;

	for (i = 0; i < htt->rx_ring.fill_cnt; i++) {
		skb = htt->rx_ring.netbufs_ring[i];
		cb = ATH10K_SKB_CB(skb);
		dma_unmap_single(htt->ar->dev, cb->paddr,
				 skb->len + skb_tailroom(skb),
				 DMA_FROM_DEVICE);
		dev_kfree_skb_any(skb);
	}

	htt->rx_ring.fill_cnt = 0;
}

static int __ath10k_htt_rx_ring_fill_n(struct ath10k_htt *htt, int num)
{
	struct htt_rx_desc *rx_desc;
	struct sk_buff *skb;
	dma_addr_t paddr;
	int ret = 0, idx;

	idx = __le32_to_cpu(*htt->rx_ring.alloc_idx.vaddr);
	while (num > 0) {
		skb = dev_alloc_skb(HTT_RX_BUF_SIZE + HTT_RX_DESC_ALIGN);
		if (!skb) {
			ret = -ENOMEM;
			goto fail;
		}

		if (!IS_ALIGNED((unsigned long)skb->data, HTT_RX_DESC_ALIGN))
			skb_pull(skb,
				 PTR_ALIGN(skb->data, HTT_RX_DESC_ALIGN) -
				 skb->data);

		/* Clear rx_desc attention word before posting to Rx ring */
		rx_desc = (struct htt_rx_desc *)skb->data;
		rx_desc->attention.flags = __cpu_to_le32(0);

		paddr = dma_map_single(htt->ar->dev, skb->data,
				       skb->len + skb_tailroom(skb),
				       DMA_FROM_DEVICE);

		if (unlikely(dma_mapping_error(htt->ar->dev, paddr))) {
			dev_kfree_skb_any(skb);
			ret = -ENOMEM;
			goto fail;
		}

		ATH10K_SKB_CB(skb)->paddr = paddr;
		htt->rx_ring.netbufs_ring[idx] = skb;
		htt->rx_ring.paddrs_ring[idx] = __cpu_to_le32(paddr);
		htt->rx_ring.fill_cnt++;

		num--;
		idx++;
		idx &= htt->rx_ring.size_mask;
	}

fail:
	*htt->rx_ring.alloc_idx.vaddr = __cpu_to_le32(idx);
	return ret;
}

static int ath10k_htt_rx_ring_fill_n(struct ath10k_htt *htt, int num)
{
	lockdep_assert_held(&htt->rx_ring.lock);
	return __ath10k_htt_rx_ring_fill_n(htt, num);
}

static void ath10k_htt_rx_msdu_buff_replenish(struct ath10k_htt *htt)
{
	int ret, num_deficit, num_to_fill;

	/* Refilling the whole RX ring buffer proves to be a bad idea. The
	 * reason is RX may take up significant amount of CPU cycles and starve
	 * other tasks, e.g. TX on an ethernet device while acting as a bridge
	 * with ath10k wlan interface. This ended up with very poor performance
	 * once CPU the host system was overwhelmed with RX on ath10k.
	 *
	 * By limiting the number of refills the replenishing occurs
	 * progressively. This in turns makes use of the fact tasklets are
	 * processed in FIFO order. This means actual RX processing can starve
	 * out refilling. If there's not enough buffers on RX ring FW will not
	 * report RX until it is refilled with enough buffers. This
	 * automatically balances load wrt to CPU power.
	 *
	 * This probably comes at a cost of lower maximum throughput but
	 * improves the avarage and stability. */
	spin_lock_bh(&htt->rx_ring.lock);
	num_deficit = htt->rx_ring.fill_level - htt->rx_ring.fill_cnt;
	num_to_fill = min(ATH10K_HTT_MAX_NUM_REFILL, num_deficit);
	num_deficit -= num_to_fill;
	ret = ath10k_htt_rx_ring_fill_n(htt, num_to_fill);
	if (ret == -ENOMEM) {
		/*
		 * Failed to fill it to the desired level -
		 * we'll start a timer and try again next time.
		 * As long as enough buffers are left in the ring for
		 * another A-MPDU rx, no special recovery is needed.
		 */
		mod_timer(&htt->rx_ring.refill_retry_timer, jiffies +
			  msecs_to_jiffies(HTT_RX_RING_REFILL_RETRY_MS));
	} else if (num_deficit > 0) {
		tasklet_schedule(&htt->rx_replenish_task);
	}
	spin_unlock_bh(&htt->rx_ring.lock);
}

static void ath10k_htt_rx_ring_refill_retry(unsigned long arg)
{
	struct ath10k_htt *htt = (struct ath10k_htt *)arg;

	ath10k_htt_rx_msdu_buff_replenish(htt);
}

static void ath10k_htt_rx_ring_clean_up(struct ath10k_htt *htt)
{
	struct sk_buff *skb;
	int i;

	for (i = 0; i < htt->rx_ring.size; i++) {
		skb = htt->rx_ring.netbufs_ring[i];
		if (!skb)
			continue;

		dma_unmap_single(htt->ar->dev, ATH10K_SKB_CB(skb)->paddr,
				 skb->len + skb_tailroom(skb),
				 DMA_FROM_DEVICE);
		dev_kfree_skb_any(skb);
		htt->rx_ring.netbufs_ring[i] = NULL;
	}
}

void ath10k_htt_rx_free(struct ath10k_htt *htt)
{
	del_timer_sync(&htt->rx_ring.refill_retry_timer);
	tasklet_kill(&htt->rx_replenish_task);
	tasklet_kill(&htt->txrx_compl_task);

	skb_queue_purge(&htt->tx_compl_q);
	skb_queue_purge(&htt->rx_compl_q);

	ath10k_htt_rx_ring_clean_up(htt);

	dma_free_coherent(htt->ar->dev,
			  (htt->rx_ring.size *
			   sizeof(htt->rx_ring.paddrs_ring)),
			  htt->rx_ring.paddrs_ring,
			  htt->rx_ring.base_paddr);

	dma_free_coherent(htt->ar->dev,
			  sizeof(*htt->rx_ring.alloc_idx.vaddr),
			  htt->rx_ring.alloc_idx.vaddr,
			  htt->rx_ring.alloc_idx.paddr);

	kfree(htt->rx_ring.netbufs_ring);
}

static inline struct sk_buff *ath10k_htt_rx_netbuf_pop(struct ath10k_htt *htt)
{
	struct ath10k *ar = htt->ar;
	int idx;
	struct sk_buff *msdu;

	lockdep_assert_held(&htt->rx_ring.lock);

	if (htt->rx_ring.fill_cnt == 0) {
		ath10k_warn(ar, "tried to pop sk_buff from an empty rx ring\n");
		return NULL;
	}

	idx = htt->rx_ring.sw_rd_idx.msdu_payld;
	msdu = htt->rx_ring.netbufs_ring[idx];
	htt->rx_ring.netbufs_ring[idx] = NULL;

	idx++;
	idx &= htt->rx_ring.size_mask;
	htt->rx_ring.sw_rd_idx.msdu_payld = idx;
	htt->rx_ring.fill_cnt--;

	dma_unmap_single(htt->ar->dev,
			 ATH10K_SKB_CB(msdu)->paddr,
			 msdu->len + skb_tailroom(msdu),
			 DMA_FROM_DEVICE);
	ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "htt rx netbuf pop: ",
			msdu->data, msdu->len + skb_tailroom(msdu));

	return msdu;
}

/* return: < 0 fatal error, 0 - non chained msdu, 1 chained msdu */
static int ath10k_htt_rx_amsdu_pop(struct ath10k_htt *htt,
				   u8 **fw_desc, int *fw_desc_len,
				   struct sk_buff_head *amsdu)
{
	struct ath10k *ar = htt->ar;
	int msdu_len, msdu_chaining = 0;
	struct sk_buff *msdu;
	struct htt_rx_desc *rx_desc;

	lockdep_assert_held(&htt->rx_ring.lock);

	for (;;) {
		int last_msdu, msdu_len_invalid, msdu_chained;

		msdu = ath10k_htt_rx_netbuf_pop(htt);
		if (!msdu) {
			__skb_queue_purge(amsdu);
			return -ENOENT;
		}

		__skb_queue_tail(amsdu, msdu);

		rx_desc = (struct htt_rx_desc *)msdu->data;

		/* FIXME: we must report msdu payload since this is what caller
		 *        expects now */
		skb_put(msdu, offsetof(struct htt_rx_desc, msdu_payload));
		skb_pull(msdu, offsetof(struct htt_rx_desc, msdu_payload));

		/*
		 * Sanity check - confirm the HW is finished filling in the
		 * rx data.
		 * If the HW and SW are working correctly, then it's guaranteed
		 * that the HW's MAC DMA is done before this point in the SW.
		 * To prevent the case that we handle a stale Rx descriptor,
		 * just assert for now until we have a way to recover.
		 */
		if (!(__le32_to_cpu(rx_desc->attention.flags)
				& RX_ATTENTION_FLAGS_MSDU_DONE)) {
			__skb_queue_purge(amsdu);
			return -EIO;
		}

		/*
		 * Copy the FW rx descriptor for this MSDU from the rx
		 * indication message into the MSDU's netbuf. HL uses the
		 * same rx indication message definition as LL, and simply
		 * appends new info (fields from the HW rx desc, and the
		 * MSDU payload itself). So, the offset into the rx
		 * indication message only has to account for the standard
		 * offset of the per-MSDU FW rx desc info within the
		 * message, and how many bytes of the per-MSDU FW rx desc
		 * info have already been consumed. (And the endianness of
		 * the host, since for a big-endian host, the rx ind
		 * message contents, including the per-MSDU rx desc bytes,
		 * were byteswapped during upload.)
		 */
		if (*fw_desc_len > 0) {
			rx_desc->fw_desc.info0 = **fw_desc;
			/*
			 * The target is expected to only provide the basic
			 * per-MSDU rx descriptors. Just to be sure, verify
			 * that the target has not attached extension data
			 * (e.g. LRO flow ID).
			 */

			/* or more, if there's extension data */
			(*fw_desc)++;
			(*fw_desc_len)--;
		} else {
			/*
			 * When an oversized AMSDU happened, FW will lost
			 * some of MSDU status - in this case, the FW
			 * descriptors provided will be less than the
			 * actual MSDUs inside this MPDU. Mark the FW
			 * descriptors so that it will still deliver to
			 * upper stack, if no CRC error for this MPDU.
			 *
			 * FIX THIS - the FW descriptors are actually for
			 * MSDUs in the end of this A-MSDU instead of the
			 * beginning.
			 */
			rx_desc->fw_desc.info0 = 0;
		}

		msdu_len_invalid = !!(__le32_to_cpu(rx_desc->attention.flags)
					& (RX_ATTENTION_FLAGS_MPDU_LENGTH_ERR |
					   RX_ATTENTION_FLAGS_MSDU_LENGTH_ERR));
		msdu_len = MS(__le32_to_cpu(rx_desc->msdu_start.info0),
			      RX_MSDU_START_INFO0_MSDU_LENGTH);
		msdu_chained = rx_desc->frag_info.ring2_more_count;

		if (msdu_len_invalid)
			msdu_len = 0;

		skb_trim(msdu, 0);
		skb_put(msdu, min(msdu_len, HTT_RX_MSDU_SIZE));
		msdu_len -= msdu->len;

		/* Note: Chained buffers do not contain rx descriptor */
		while (msdu_chained--) {
			msdu = ath10k_htt_rx_netbuf_pop(htt);
			if (!msdu) {
				__skb_queue_purge(amsdu);
				return -ENOENT;
			}

			__skb_queue_tail(amsdu, msdu);
			skb_trim(msdu, 0);
			skb_put(msdu, min(msdu_len, HTT_RX_BUF_SIZE));
			msdu_len -= msdu->len;
			msdu_chaining = 1;
		}

		last_msdu = __le32_to_cpu(rx_desc->msdu_end.info0) &
				RX_MSDU_END_INFO0_LAST_MSDU;

		trace_ath10k_htt_rx_desc(ar, &rx_desc->attention,
					 sizeof(*rx_desc) - sizeof(u32));

		if (last_msdu)
			break;
	}

	if (skb_queue_empty(amsdu))
		msdu_chaining = -1;

	/*
	 * Don't refill the ring yet.
	 *
	 * First, the elements popped here are still in use - it is not
	 * safe to overwrite them until the matching call to
	 * mpdu_desc_list_next. Second, for efficiency it is preferable to
	 * refill the rx ring with 1 PPDU's worth of rx buffers (something
	 * like 32 x 3 buffers), rather than one MPDU's worth of rx buffers
	 * (something like 3 buffers). Consequently, we'll rely on the txrx
	 * SW to tell us when it is done pulling all the PPDU's rx buffers
	 * out of the rx ring, and then refill it just once.
	 */

	return msdu_chaining;
}

static void ath10k_htt_rx_replenish_task(unsigned long ptr)
{
	struct ath10k_htt *htt = (struct ath10k_htt *)ptr;

	ath10k_htt_rx_msdu_buff_replenish(htt);
}

int ath10k_htt_rx_alloc(struct ath10k_htt *htt)
{
	struct ath10k *ar = htt->ar;
	dma_addr_t paddr;
	void *vaddr;
	size_t size;
	struct timer_list *timer = &htt->rx_ring.refill_retry_timer;

	htt->rx_confused = false;

	htt->rx_ring.size = ath10k_htt_rx_ring_size(htt);
	if (!is_power_of_2(htt->rx_ring.size)) {
		ath10k_warn(ar, "htt rx ring size is not power of 2\n");
		return -EINVAL;
	}

	htt->rx_ring.size_mask = htt->rx_ring.size - 1;

	/*
	 * Set the initial value for the level to which the rx ring
	 * should be filled, based on the max throughput and the
	 * worst likely latency for the host to fill the rx ring
	 * with new buffers. In theory, this fill level can be
	 * dynamically adjusted from the initial value set here, to
	 * reflect the actual host latency rather than a
	 * conservative assumption about the host latency.
	 */
	htt->rx_ring.fill_level = ath10k_htt_rx_ring_fill_level(htt);

	htt->rx_ring.netbufs_ring =
		kzalloc(htt->rx_ring.size * sizeof(struct sk_buff *),
			GFP_KERNEL);
	if (!htt->rx_ring.netbufs_ring)
		goto err_netbuf;

	size = htt->rx_ring.size * sizeof(htt->rx_ring.paddrs_ring);

	vaddr = dma_alloc_coherent(htt->ar->dev, size, &paddr, GFP_DMA);
	if (!vaddr)
		goto err_dma_ring;

	htt->rx_ring.paddrs_ring = vaddr;
	htt->rx_ring.base_paddr = paddr;

	vaddr = dma_alloc_coherent(htt->ar->dev,
				   sizeof(*htt->rx_ring.alloc_idx.vaddr),
				   &paddr, GFP_DMA);
	if (!vaddr)
		goto err_dma_idx;

	htt->rx_ring.alloc_idx.vaddr = vaddr;
	htt->rx_ring.alloc_idx.paddr = paddr;
	htt->rx_ring.sw_rd_idx.msdu_payld = 0;
	*htt->rx_ring.alloc_idx.vaddr = 0;

	/* Initialize the Rx refill retry timer */
	setup_timer(timer, ath10k_htt_rx_ring_refill_retry, (unsigned long)htt);

	spin_lock_init(&htt->rx_ring.lock);

	htt->rx_ring.fill_cnt = 0;
	if (__ath10k_htt_rx_ring_fill_n(htt, htt->rx_ring.fill_level))
		goto err_fill_ring;

	tasklet_init(&htt->rx_replenish_task, ath10k_htt_rx_replenish_task,
		     (unsigned long)htt);

	skb_queue_head_init(&htt->tx_compl_q);
	skb_queue_head_init(&htt->rx_compl_q);

	tasklet_init(&htt->txrx_compl_task, ath10k_htt_txrx_compl_task,
		     (unsigned long)htt);

	ath10k_dbg(ar, ATH10K_DBG_BOOT, "htt rx ring size %d fill_level %d\n",
		   htt->rx_ring.size, htt->rx_ring.fill_level);
	return 0;

err_fill_ring:
	ath10k_htt_rx_ring_free(htt);
	dma_free_coherent(htt->ar->dev,
			  sizeof(*htt->rx_ring.alloc_idx.vaddr),
			  htt->rx_ring.alloc_idx.vaddr,
			  htt->rx_ring.alloc_idx.paddr);
err_dma_idx:
	dma_free_coherent(htt->ar->dev,
			  (htt->rx_ring.size *
			   sizeof(htt->rx_ring.paddrs_ring)),
			  htt->rx_ring.paddrs_ring,
			  htt->rx_ring.base_paddr);
err_dma_ring:
	kfree(htt->rx_ring.netbufs_ring);
err_netbuf:
	return -ENOMEM;
}

static int ath10k_htt_rx_crypto_param_len(struct ath10k *ar,
					  enum htt_rx_mpdu_encrypt_type type)
{
	switch (type) {
	case HTT_RX_MPDU_ENCRYPT_NONE:
		return 0;
	case HTT_RX_MPDU_ENCRYPT_WEP40:
	case HTT_RX_MPDU_ENCRYPT_WEP104:
		return IEEE80211_WEP_IV_LEN;
	case HTT_RX_MPDU_ENCRYPT_TKIP_WITHOUT_MIC:
	case HTT_RX_MPDU_ENCRYPT_TKIP_WPA:
		return IEEE80211_TKIP_IV_LEN;
	case HTT_RX_MPDU_ENCRYPT_AES_CCM_WPA2:
		return IEEE80211_CCMP_HDR_LEN;
	case HTT_RX_MPDU_ENCRYPT_WEP128:
	case HTT_RX_MPDU_ENCRYPT_WAPI:
		break;
	}

	ath10k_warn(ar, "unsupported encryption type %d\n", type);
	return 0;
}

#define MICHAEL_MIC_LEN 8

static int ath10k_htt_rx_crypto_tail_len(struct ath10k *ar,
					 enum htt_rx_mpdu_encrypt_type type)
{
	switch (type) {
	case HTT_RX_MPDU_ENCRYPT_NONE:
		return 0;
	case HTT_RX_MPDU_ENCRYPT_WEP40:
	case HTT_RX_MPDU_ENCRYPT_WEP104:
		return IEEE80211_WEP_ICV_LEN;
	case HTT_RX_MPDU_ENCRYPT_TKIP_WITHOUT_MIC:
	case HTT_RX_MPDU_ENCRYPT_TKIP_WPA:
		return IEEE80211_TKIP_ICV_LEN;
	case HTT_RX_MPDU_ENCRYPT_AES_CCM_WPA2:
		return IEEE80211_CCMP_MIC_LEN;
	case HTT_RX_MPDU_ENCRYPT_WEP128:
	case HTT_RX_MPDU_ENCRYPT_WAPI:
		break;
	}

	ath10k_warn(ar, "unsupported encryption type %d\n", type);
	return 0;
}

struct rfc1042_hdr {
	u8 llc_dsap;
	u8 llc_ssap;
	u8 llc_ctrl;
	u8 snap_oui[3];
	__be16 snap_type;
} __packed;

struct amsdu_subframe_hdr {
	u8 dst[ETH_ALEN];
	u8 src[ETH_ALEN];
	__be16 len;
} __packed;

static const u8 rx_legacy_rate_idx[] = {
	3,	/* 0x00  - 11Mbps  */
	2,	/* 0x01  - 5.5Mbps */
	1,	/* 0x02  - 2Mbps   */
	0,	/* 0x03  - 1Mbps   */
	3,	/* 0x04  - 11Mbps  */
	2,	/* 0x05  - 5.5Mbps */
	1,	/* 0x06  - 2Mbps   */
	0,	/* 0x07  - 1Mbps   */
	10,	/* 0x08  - 48Mbps  */
	8,	/* 0x09  - 24Mbps  */
	6,	/* 0x0A  - 12Mbps  */
	4,	/* 0x0B  - 6Mbps   */
	11,	/* 0x0C  - 54Mbps  */
	9,	/* 0x0D  - 36Mbps  */
	7,	/* 0x0E  - 18Mbps  */
	5,	/* 0x0F  - 9Mbps   */
};

static void ath10k_htt_rx_h_rates(struct ath10k *ar,
				  struct ieee80211_rx_status *status,
				  struct htt_rx_desc *rxd)
{
	enum ieee80211_band band;
	u8 cck, rate, rate_idx, bw, sgi, mcs, nss;
	u8 preamble = 0;
	u32 info1, info2, info3;

	/* Band value can't be set as undefined but freq can be 0 - use that to
	 * determine whether band is provided.
	 *
	 * FIXME: Perhaps this can go away if CCK rate reporting is a little
	 * reworked?
	 */
	if (!status->freq)
		return;

	band = status->band;
	info1 = __le32_to_cpu(rxd->ppdu_start.info1);
	info2 = __le32_to_cpu(rxd->ppdu_start.info2);
	info3 = __le32_to_cpu(rxd->ppdu_start.info3);

	preamble = MS(info1, RX_PPDU_START_INFO1_PREAMBLE_TYPE);

	switch (preamble) {
	case HTT_RX_LEGACY:
		cck = info1 & RX_PPDU_START_INFO1_L_SIG_RATE_SELECT;
		rate = MS(info1, RX_PPDU_START_INFO1_L_SIG_RATE);
		rate_idx = 0;

		if (rate < 0x08 || rate > 0x0F)
			break;

		switch (band) {
		case IEEE80211_BAND_2GHZ:
			if (cck)
				rate &= ~BIT(3);
			rate_idx = rx_legacy_rate_idx[rate];
			break;
		case IEEE80211_BAND_5GHZ:
			rate_idx = rx_legacy_rate_idx[rate];
			/* We are using same rate table registering
			   HW - ath10k_rates[]. In case of 5GHz skip
			   CCK rates, so -4 here */
			rate_idx -= 4;
			break;
		default:
			break;
		}

		status->rate_idx = rate_idx;
		break;
	case HTT_RX_HT:
	case HTT_RX_HT_WITH_TXBF:
		/* HT-SIG - Table 20-11 in info2 and info3 */
		mcs = info2 & 0x1F;
		nss = mcs >> 3;
		bw = (info2 >> 7) & 1;
		sgi = (info3 >> 7) & 1;

		status->rate_idx = mcs;
		status->flag |= RX_FLAG_HT;
		if (sgi)
			status->flag |= RX_FLAG_SHORT_GI;
		if (bw)
			status->flag |= RX_FLAG_40MHZ;
		break;
	case HTT_RX_VHT:
	case HTT_RX_VHT_WITH_TXBF:
		/* VHT-SIG-A1 in info2, VHT-SIG-A2 in info3
		   TODO check this */
		mcs = (info3 >> 4) & 0x0F;
		nss = ((info2 >> 10) & 0x07) + 1;
		bw = info2 & 3;
		sgi = info3 & 1;

		status->rate_idx = mcs;
		status->vht_nss = nss;

		if (sgi)
			status->flag |= RX_FLAG_SHORT_GI;

		switch (bw) {
		/* 20MHZ */
		case 0:
			break;
		/* 40MHZ */
		case 1:
			status->flag |= RX_FLAG_40MHZ;
			break;
		/* 80MHZ */
		case 2:
			status->vht_flag |= RX_VHT_FLAG_80MHZ;
		}

		status->flag |= RX_FLAG_VHT;
		break;
	default:
		break;
	}
}

static bool ath10k_htt_rx_h_channel(struct ath10k *ar,
				    struct ieee80211_rx_status *status)
{
	struct ieee80211_channel *ch;

	spin_lock_bh(&ar->data_lock);
	ch = ar->scan_channel;
	if (!ch)
		ch = ar->rx_channel;
	spin_unlock_bh(&ar->data_lock);

	if (!ch)
		return false;

	status->band = ch->band;
	status->freq = ch->center_freq;

	return true;
}

static void ath10k_htt_rx_h_signal(struct ath10k *ar,
				   struct ieee80211_rx_status *status,
				   struct htt_rx_desc *rxd)
{
	/* FIXME: Get real NF */
	status->signal = ATH10K_DEFAULT_NOISE_FLOOR +
			 rxd->ppdu_start.rssi_comb;
	status->flag &= ~RX_FLAG_NO_SIGNAL_VAL;
}

static void ath10k_htt_rx_h_mactime(struct ath10k *ar,
				    struct ieee80211_rx_status *status,
				    struct htt_rx_desc *rxd)
{
	/* FIXME: TSF is known only at the end of PPDU, in the last MPDU. This
	 * means all prior MSDUs in a PPDU are reported to mac80211 without the
	 * TSF. Is it worth holding frames until end of PPDU is known?
	 *
	 * FIXME: Can we get/compute 64bit TSF?
	 */
	status->mactime = __le32_to_cpu(rxd->ppdu_end.tsf_timestamp);
	status->flag |= RX_FLAG_MACTIME_END;
}

static void ath10k_htt_rx_h_ppdu(struct ath10k *ar,
				 struct sk_buff_head *amsdu,
				 struct ieee80211_rx_status *status)
{
	struct sk_buff *first;
	struct htt_rx_desc *rxd;
	bool is_first_ppdu;
	bool is_last_ppdu;

	if (skb_queue_empty(amsdu))
		return;

	first = skb_peek(amsdu);
	rxd = (void *)first->data - sizeof(*rxd);

	is_first_ppdu = !!(rxd->attention.flags &
			   __cpu_to_le32(RX_ATTENTION_FLAGS_FIRST_MPDU));
	is_last_ppdu = !!(rxd->attention.flags &
			  __cpu_to_le32(RX_ATTENTION_FLAGS_LAST_MPDU));

	if (is_first_ppdu) {
		/* New PPDU starts so clear out the old per-PPDU status. */
		status->freq = 0;
		status->rate_idx = 0;
		status->vht_nss = 0;
		status->vht_flag &= ~RX_VHT_FLAG_80MHZ;
		status->flag &= ~(RX_FLAG_HT |
				  RX_FLAG_VHT |
				  RX_FLAG_SHORT_GI |
				  RX_FLAG_40MHZ |
				  RX_FLAG_MACTIME_END);
		status->flag |= RX_FLAG_NO_SIGNAL_VAL;

		ath10k_htt_rx_h_signal(ar, status, rxd);
		ath10k_htt_rx_h_channel(ar, status);
		ath10k_htt_rx_h_rates(ar, status, rxd);
	}

	if (is_last_ppdu)
		ath10k_htt_rx_h_mactime(ar, status, rxd);
}

static const char * const tid_to_ac[] = {
	"BE",
	"BK",
	"BK",
	"BE",
	"VI",
	"VI",
	"VO",
	"VO",
};

static char *ath10k_get_tid(struct ieee80211_hdr *hdr, char *out, size_t size)
{
	u8 *qc;
	int tid;

	if (!ieee80211_is_data_qos(hdr->frame_control))
		return "";

	qc = ieee80211_get_qos_ctl(hdr);
	tid = *qc & IEEE80211_QOS_CTL_TID_MASK;
	if (tid < 8)
		snprintf(out, size, "tid %d (%s)", tid, tid_to_ac[tid]);
	else
		snprintf(out, size, "tid %d", tid);

	return out;
}

static void ath10k_process_rx(struct ath10k *ar,
			      struct ieee80211_rx_status *rx_status,
			      struct sk_buff *skb)
{
	struct ieee80211_rx_status *status;
	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data;
	char tid[32];

	status = IEEE80211_SKB_RXCB(skb);
	*status = *rx_status;

	ath10k_dbg(ar, ATH10K_DBG_DATA,
		   "rx skb %p len %u peer %pM %s %s sn %u %s%s%s%s%s %srate_idx %u vht_nss %u freq %u band %u flag 0x%x fcs-err %i mic-err %i amsdu-more %i\n",
		   skb,
		   skb->len,
		   ieee80211_get_SA(hdr),
		   ath10k_get_tid(hdr, tid, sizeof(tid)),
		   is_multicast_ether_addr(ieee80211_get_DA(hdr)) ?
							"mcast" : "ucast",
		   (__le16_to_cpu(hdr->seq_ctrl) & IEEE80211_SCTL_SEQ) >> 4,
		   status->flag == 0 ? "legacy" : "",
		   status->flag & RX_FLAG_HT ? "ht" : "",
		   status->flag & RX_FLAG_VHT ? "vht" : "",
		   status->flag & RX_FLAG_40MHZ ? "40" : "",
		   status->vht_flag & RX_VHT_FLAG_80MHZ ? "80" : "",
		   status->flag & RX_FLAG_SHORT_GI ? "sgi " : "",
		   status->rate_idx,
		   status->vht_nss,
		   status->freq,
		   status->band, status->flag,
		   !!(status->flag & RX_FLAG_FAILED_FCS_CRC),
		   !!(status->flag & RX_FLAG_MMIC_ERROR),
		   !!(status->flag & RX_FLAG_AMSDU_MORE));
	ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "rx skb: ",
			skb->data, skb->len);
	trace_ath10k_rx_hdr(ar, skb->data, skb->len);
	trace_ath10k_rx_payload(ar, skb->data, skb->len);

	ieee80211_rx(ar->hw, skb);
}

static int ath10k_htt_rx_nwifi_hdrlen(struct ieee80211_hdr *hdr)
{
	/* nwifi header is padded to 4 bytes. this fixes 4addr rx */
	return round_up(ieee80211_hdrlen(hdr->frame_control), 4);
}

static void ath10k_htt_rx_h_undecap_raw(struct ath10k *ar,
					struct sk_buff *msdu,
					struct ieee80211_rx_status *status,
					enum htt_rx_mpdu_encrypt_type enctype,
					bool is_decrypted)
{
	struct ieee80211_hdr *hdr;
	struct htt_rx_desc *rxd;
	size_t hdr_len;
	size_t crypto_len;
	bool is_first;
	bool is_last;

	rxd = (void *)msdu->data - sizeof(*rxd);
	is_first = !!(rxd->msdu_end.info0 &
		      __cpu_to_le32(RX_MSDU_END_INFO0_FIRST_MSDU));
	is_last = !!(rxd->msdu_end.info0 &
		     __cpu_to_le32(RX_MSDU_END_INFO0_LAST_MSDU));

	/* Delivered decapped frame:
	 * [802.11 header]
	 * [crypto param] <-- can be trimmed if !fcs_err &&
	 *                    !decrypt_err && !peer_idx_invalid
	 * [amsdu header] <-- only if A-MSDU
	 * [rfc1042/llc]
	 * [payload]
	 * [FCS] <-- at end, needs to be trimmed
	 */

	/* This probably shouldn't happen but warn just in case */
	if (unlikely(WARN_ON_ONCE(!is_first)))
		return;

	/* This probably shouldn't happen but warn just in case */
	if (unlikely(WARN_ON_ONCE(!(is_first && is_last))))
		return;

	skb_trim(msdu, msdu->len - FCS_LEN);

	/* In most cases this will be true for sniffed frames. It makes sense
	 * to deliver them as-is without stripping the crypto param. This would
	 * also make sense for software based decryption (which is not
	 * implemented in ath10k).
	 *
	 * If there's no error then the frame is decrypted. At least that is
	 * the case for frames that come in via fragmented rx indication.
	 */
	if (!is_decrypted)
		return;

	/* The payload is decrypted so strip crypto params. Start from tail
	 * since hdr is used to compute some stuff.
	 */

	hdr = (void *)msdu->data;

	/* Tail */
	skb_trim(msdu, msdu->len - ath10k_htt_rx_crypto_tail_len(ar, enctype));

	/* MMIC */
	if (!ieee80211_has_morefrags(hdr->frame_control) &&
	    enctype == HTT_RX_MPDU_ENCRYPT_TKIP_WPA)
		skb_trim(msdu, msdu->len - 8);

	/* Head */
	hdr_len = ieee80211_hdrlen(hdr->frame_control);
	crypto_len = ath10k_htt_rx_crypto_param_len(ar, enctype);

	memmove((void *)msdu->data + crypto_len,
		(void *)msdu->data, hdr_len);
	skb_pull(msdu, crypto_len);
}

static void ath10k_htt_rx_h_undecap_nwifi(struct ath10k *ar,
					  struct sk_buff *msdu,
					  struct ieee80211_rx_status *status,
					  const u8 first_hdr[64])
{
	struct ieee80211_hdr *hdr;
	size_t hdr_len;
	u8 da[ETH_ALEN];
	u8 sa[ETH_ALEN];

	/* Delivered decapped frame:
	 * [nwifi 802.11 header] <-- replaced with 802.11 hdr
	 * [rfc1042/llc]
	 *
	 * Note: The nwifi header doesn't have QoS Control and is
	 * (always?) a 3addr frame.
	 *
	 * Note2: There's no A-MSDU subframe header. Even if it's part
	 * of an A-MSDU.
	 */

	/* pull decapped header and copy SA & DA */
	hdr = (struct ieee80211_hdr *)msdu->data;
	hdr_len = ath10k_htt_rx_nwifi_hdrlen(hdr);
	ether_addr_copy(da, ieee80211_get_DA(hdr));
	ether_addr_copy(sa, ieee80211_get_SA(hdr));
	skb_pull(msdu, hdr_len);

	/* push original 802.11 header */
	hdr = (struct ieee80211_hdr *)first_hdr;
	hdr_len = ieee80211_hdrlen(hdr->frame_control);
	memcpy(skb_push(msdu, hdr_len), hdr, hdr_len);

	/* original 802.11 header has a different DA and in
	 * case of 4addr it may also have different SA
	 */
	hdr = (struct ieee80211_hdr *)msdu->data;
	ether_addr_copy(ieee80211_get_DA(hdr), da);
	ether_addr_copy(ieee80211_get_SA(hdr), sa);
}

static void *ath10k_htt_rx_h_find_rfc1042(struct ath10k *ar,
					  struct sk_buff *msdu,
					  enum htt_rx_mpdu_encrypt_type enctype)
{
	struct ieee80211_hdr *hdr;
	struct htt_rx_desc *rxd;
	size_t hdr_len, crypto_len;
	void *rfc1042;
	bool is_first, is_last, is_amsdu;

	rxd = (void *)msdu->data - sizeof(*rxd);
	hdr = (void *)rxd->rx_hdr_status;

	is_first = !!(rxd->msdu_end.info0 &
		      __cpu_to_le32(RX_MSDU_END_INFO0_FIRST_MSDU));
	is_last = !!(rxd->msdu_end.info0 &
		     __cpu_to_le32(RX_MSDU_END_INFO0_LAST_MSDU));
	is_amsdu = !(is_first && is_last);

	rfc1042 = hdr;

	if (is_first) {
		hdr_len = ieee80211_hdrlen(hdr->frame_control);
		crypto_len = ath10k_htt_rx_crypto_param_len(ar, enctype);

		rfc1042 += round_up(hdr_len, 4) +
			   round_up(crypto_len, 4);
	}

	if (is_amsdu)
		rfc1042 += sizeof(struct amsdu_subframe_hdr);

	return rfc1042;
}

static void ath10k_htt_rx_h_undecap_eth(struct ath10k *ar,
					struct sk_buff *msdu,
					struct ieee80211_rx_status *status,
					const u8 first_hdr[64],
					enum htt_rx_mpdu_encrypt_type enctype)
{
	struct ieee80211_hdr *hdr;
	struct ethhdr *eth;
	size_t hdr_len;
	void *rfc1042;
	u8 da[ETH_ALEN];
	u8 sa[ETH_ALEN];

	/* Delivered decapped frame:
	 * [eth header] <-- replaced with 802.11 hdr & rfc1042/llc
	 * [payload]
	 */

	rfc1042 = ath10k_htt_rx_h_find_rfc1042(ar, msdu, enctype);
	if (WARN_ON_ONCE(!rfc1042))
		return;

	/* pull decapped header and copy SA & DA */
	eth = (struct ethhdr *)msdu->data;
	ether_addr_copy(da, eth->h_dest);
	ether_addr_copy(sa, eth->h_source);
	skb_pull(msdu, sizeof(struct ethhdr));

	/* push rfc1042/llc/snap */
	memcpy(skb_push(msdu, sizeof(struct rfc1042_hdr)), rfc1042,
	       sizeof(struct rfc1042_hdr));

	/* push original 802.11 header */
	hdr = (struct ieee80211_hdr *)first_hdr;
	hdr_len = ieee80211_hdrlen(hdr->frame_control);
	memcpy(skb_push(msdu, hdr_len), hdr, hdr_len);

	/* original 802.11 header has a different DA and in
	 * case of 4addr it may also have different SA
	 */
	hdr = (struct ieee80211_hdr *)msdu->data;
	ether_addr_copy(ieee80211_get_DA(hdr), da);
	ether_addr_copy(ieee80211_get_SA(hdr), sa);
}

static void ath10k_htt_rx_h_undecap_snap(struct ath10k *ar,
					 struct sk_buff *msdu,
					 struct ieee80211_rx_status *status,
					 const u8 first_hdr[64])
{
	struct ieee80211_hdr *hdr;
	size_t hdr_len;

	/* Delivered decapped frame:
	 * [amsdu header] <-- replaced with 802.11 hdr
	 * [rfc1042/llc]
	 * [payload]
	 */

	skb_pull(msdu, sizeof(struct amsdu_subframe_hdr));

	hdr = (struct ieee80211_hdr *)first_hdr;
	hdr_len = ieee80211_hdrlen(hdr->frame_control);
	memcpy(skb_push(msdu, hdr_len), hdr, hdr_len);
}

static void ath10k_htt_rx_h_undecap(struct ath10k *ar,
				    struct sk_buff *msdu,
				    struct ieee80211_rx_status *status,
				    u8 first_hdr[64],
				    enum htt_rx_mpdu_encrypt_type enctype,
				    bool is_decrypted)
{
	struct htt_rx_desc *rxd;
	enum rx_msdu_decap_format decap;
	struct ieee80211_hdr *hdr;

	/* First msdu's decapped header:
	 * [802.11 header] <-- padded to 4 bytes long
	 * [crypto param] <-- padded to 4 bytes long
	 * [amsdu header] <-- only if A-MSDU
	 * [rfc1042/llc]
	 *
	 * Other (2nd, 3rd, ..) msdu's decapped header:
	 * [amsdu header] <-- only if A-MSDU
	 * [rfc1042/llc]
	 */

	rxd = (void *)msdu->data - sizeof(*rxd);
	hdr = (void *)rxd->rx_hdr_status;
	decap = MS(__le32_to_cpu(rxd->msdu_start.info1),
		   RX_MSDU_START_INFO1_DECAP_FORMAT);

	switch (decap) {
	case RX_MSDU_DECAP_RAW:
		ath10k_htt_rx_h_undecap_raw(ar, msdu, status, enctype,
					    is_decrypted);
		break;
	case RX_MSDU_DECAP_NATIVE_WIFI:
		ath10k_htt_rx_h_undecap_nwifi(ar, msdu, status, first_hdr);
		break;
	case RX_MSDU_DECAP_ETHERNET2_DIX:
		ath10k_htt_rx_h_undecap_eth(ar, msdu, status, first_hdr, enctype);
		break;
	case RX_MSDU_DECAP_8023_SNAP_LLC:
		ath10k_htt_rx_h_undecap_snap(ar, msdu, status, first_hdr);
		break;
	}
}

static int ath10k_htt_rx_get_csum_state(struct sk_buff *skb)
{
	struct htt_rx_desc *rxd;
	u32 flags, info;
	bool is_ip4, is_ip6;
	bool is_tcp, is_udp;
	bool ip_csum_ok, tcpudp_csum_ok;

	rxd = (void *)skb->data - sizeof(*rxd);
	flags = __le32_to_cpu(rxd->attention.flags);
	info = __le32_to_cpu(rxd->msdu_start.info1);

	is_ip4 = !!(info & RX_MSDU_START_INFO1_IPV4_PROTO);
	is_ip6 = !!(info & RX_MSDU_START_INFO1_IPV6_PROTO);
	is_tcp = !!(info & RX_MSDU_START_INFO1_TCP_PROTO);
	is_udp = !!(info & RX_MSDU_START_INFO1_UDP_PROTO);
	ip_csum_ok = !(flags & RX_ATTENTION_FLAGS_IP_CHKSUM_FAIL);
	tcpudp_csum_ok = !(flags & RX_ATTENTION_FLAGS_TCP_UDP_CHKSUM_FAIL);

	if (!is_ip4 && !is_ip6)
		return CHECKSUM_NONE;
	if (!is_tcp && !is_udp)
		return CHECKSUM_NONE;
	if (!ip_csum_ok)
		return CHECKSUM_NONE;
	if (!tcpudp_csum_ok)
		return CHECKSUM_NONE;

	return CHECKSUM_UNNECESSARY;
}

static void ath10k_htt_rx_h_csum_offload(struct sk_buff *msdu)
{
	msdu->ip_summed = ath10k_htt_rx_get_csum_state(msdu);
}

static void ath10k_htt_rx_h_mpdu(struct ath10k *ar,
				 struct sk_buff_head *amsdu,
				 struct ieee80211_rx_status *status)
{
	struct sk_buff *first;
	struct sk_buff *last;
	struct sk_buff *msdu;
	struct htt_rx_desc *rxd;
	struct ieee80211_hdr *hdr;
	enum htt_rx_mpdu_encrypt_type enctype;
	u8 first_hdr[64];
	u8 *qos;
	size_t hdr_len;
	bool has_fcs_err;
	bool has_crypto_err;
	bool has_tkip_err;
	bool has_peer_idx_invalid;
	bool is_decrypted;
	u32 attention;

	if (skb_queue_empty(amsdu))
		return;

	first = skb_peek(amsdu);
	rxd = (void *)first->data - sizeof(*rxd);

	enctype = MS(__le32_to_cpu(rxd->mpdu_start.info0),
		     RX_MPDU_START_INFO0_ENCRYPT_TYPE);

	/* First MSDU's Rx descriptor in an A-MSDU contains full 802.11
	 * decapped header. It'll be used for undecapping of each MSDU.
	 */
	hdr = (void *)rxd->rx_hdr_status;
	hdr_len = ieee80211_hdrlen(hdr->frame_control);
	memcpy(first_hdr, hdr, hdr_len);

	/* Each A-MSDU subframe will use the original header as the base and be
	 * reported as a separate MSDU so strip the A-MSDU bit from QoS Ctl.
	 */
	hdr = (void *)first_hdr;
	qos = ieee80211_get_qos_ctl(hdr);
	qos[0] &= ~IEEE80211_QOS_CTL_A_MSDU_PRESENT;

	/* Some attention flags are valid only in the last MSDU. */
	last = skb_peek_tail(amsdu);
	rxd = (void *)last->data - sizeof(*rxd);
	attention = __le32_to_cpu(rxd->attention.flags);

	has_fcs_err = !!(attention & RX_ATTENTION_FLAGS_FCS_ERR);
	has_crypto_err = !!(attention & RX_ATTENTION_FLAGS_DECRYPT_ERR);
	has_tkip_err = !!(attention & RX_ATTENTION_FLAGS_TKIP_MIC_ERR);
	has_peer_idx_invalid = !!(attention & RX_ATTENTION_FLAGS_PEER_IDX_INVALID);

	/* Note: If hardware captures an encrypted frame that it can't decrypt,
	 * e.g. due to fcs error, missing peer or invalid key data it will
	 * report the frame as raw.
	 */
	is_decrypted = (enctype != HTT_RX_MPDU_ENCRYPT_NONE &&
			!has_fcs_err &&
			!has_crypto_err &&
			!has_peer_idx_invalid);

	/* Clear per-MPDU flags while leaving per-PPDU flags intact. */
	status->flag &= ~(RX_FLAG_FAILED_FCS_CRC |
			  RX_FLAG_MMIC_ERROR |
			  RX_FLAG_DECRYPTED |
			  RX_FLAG_IV_STRIPPED |
			  RX_FLAG_MMIC_STRIPPED);

	if (has_fcs_err)
		status->flag |= RX_FLAG_FAILED_FCS_CRC;

	if (has_tkip_err)
		status->flag |= RX_FLAG_MMIC_ERROR;

	if (is_decrypted)
		status->flag |= RX_FLAG_DECRYPTED |
				RX_FLAG_IV_STRIPPED |
				RX_FLAG_MMIC_STRIPPED;

	skb_queue_walk(amsdu, msdu) {
		ath10k_htt_rx_h_csum_offload(msdu);
		ath10k_htt_rx_h_undecap(ar, msdu, status, first_hdr, enctype,
					is_decrypted);

		/* Undecapping involves copying the original 802.11 header back
		 * to sk_buff. If frame is protected and hardware has decrypted
		 * it then remove the protected bit.
		 */
		if (!is_decrypted)
			continue;

		hdr = (void *)msdu->data;
		hdr->frame_control &= ~__cpu_to_le16(IEEE80211_FCTL_PROTECTED);
	}
}

static void ath10k_htt_rx_h_deliver(struct ath10k *ar,
				    struct sk_buff_head *amsdu,
				    struct ieee80211_rx_status *status)
{
	struct sk_buff *msdu;

	while ((msdu = __skb_dequeue(amsdu))) {
		/* Setup per-MSDU flags */
		if (skb_queue_empty(amsdu))
			status->flag &= ~RX_FLAG_AMSDU_MORE;
		else
			status->flag |= RX_FLAG_AMSDU_MORE;

		ath10k_process_rx(ar, status, msdu);
	}
}

static int ath10k_unchain_msdu(struct sk_buff_head *amsdu)
{
	struct sk_buff *skb, *first;
	int space;
	int total_len = 0;

	/* TODO:  Might could optimize this by using
	 * skb_try_coalesce or similar method to
	 * decrease copying, or maybe get mac80211 to
	 * provide a way to just receive a list of
	 * skb?
	 */

	first = __skb_dequeue(amsdu);

	/* Allocate total length all at once. */
	skb_queue_walk(amsdu, skb)
		total_len += skb->len;

	space = total_len - skb_tailroom(first);
	if ((space > 0) &&
	    (pskb_expand_head(first, 0, space, GFP_ATOMIC) < 0)) {
		/* TODO:  bump some rx-oom error stat */
		/* put it back together so we can free the
		 * whole list at once.
		 */
		__skb_queue_head(amsdu, first);
		return -1;
	}

	/* Walk list again, copying contents into
	 * msdu_head
	 */
	while ((skb = __skb_dequeue(amsdu))) {
		skb_copy_from_linear_data(skb, skb_put(first, skb->len),
					  skb->len);
		dev_kfree_skb_any(skb);
	}

	__skb_queue_head(amsdu, first);
	return 0;
}

static void ath10k_htt_rx_h_unchain(struct ath10k *ar,
				    struct sk_buff_head *amsdu,
				    bool chained)
{
	struct sk_buff *first;
	struct htt_rx_desc *rxd;
	enum rx_msdu_decap_format decap;

	first = skb_peek(amsdu);
	rxd = (void *)first->data - sizeof(*rxd);
	decap = MS(__le32_to_cpu(rxd->msdu_start.info1),
		   RX_MSDU_START_INFO1_DECAP_FORMAT);

	if (!chained)
		return;

	/* FIXME: Current unchaining logic can only handle simple case of raw
	 * msdu chaining. If decapping is other than raw the chaining may be
	 * more complex and this isn't handled by the current code. Don't even
	 * try re-constructing such frames - it'll be pretty much garbage.
	 */
	if (decap != RX_MSDU_DECAP_RAW ||
	    skb_queue_len(amsdu) != 1 + rxd->frag_info.ring2_more_count) {
		__skb_queue_purge(amsdu);
		return;
	}

	ath10k_unchain_msdu(amsdu);
}

static bool ath10k_htt_rx_amsdu_allowed(struct ath10k *ar,
					struct sk_buff_head *amsdu,
					struct ieee80211_rx_status *rx_status)
{
	struct sk_buff *msdu;
	struct htt_rx_desc *rxd;
	bool is_mgmt;
	bool has_fcs_err;

	msdu = skb_peek(amsdu);
	rxd = (void *)msdu->data - sizeof(*rxd);

	/* FIXME: It might be a good idea to do some fuzzy-testing to drop
	 * invalid/dangerous frames.
	 */

	if (!rx_status->freq) {
		ath10k_warn(ar, "no channel configured; ignoring frame(s)!\n");
		return false;
	}

	is_mgmt = !!(rxd->attention.flags &
		     __cpu_to_le32(RX_ATTENTION_FLAGS_MGMT_TYPE));
	has_fcs_err = !!(rxd->attention.flags &
			 __cpu_to_le32(RX_ATTENTION_FLAGS_FCS_ERR));

	/* Management frames are handled via WMI events. The pros of such
	 * approach is that channel is explicitly provided in WMI events
	 * whereas HTT doesn't provide channel information for Rxed frames.
	 *
	 * However some firmware revisions don't report corrupted frames via
	 * WMI so don't drop them.
	 */
	if (is_mgmt && !has_fcs_err) {
		ath10k_dbg(ar, ATH10K_DBG_HTT, "htt rx mgmt ctrl\n");
		return false;
	}

	if (test_bit(ATH10K_CAC_RUNNING, &ar->dev_flags)) {
		ath10k_dbg(ar, ATH10K_DBG_HTT, "htt rx cac running\n");
		return false;
	}

	return true;
}

static void ath10k_htt_rx_h_filter(struct ath10k *ar,
				   struct sk_buff_head *amsdu,
				   struct ieee80211_rx_status *rx_status)
{
	if (skb_queue_empty(amsdu))
		return;

	if (ath10k_htt_rx_amsdu_allowed(ar, amsdu, rx_status))
		return;

	__skb_queue_purge(amsdu);
}

static void ath10k_htt_rx_handler(struct ath10k_htt *htt,
				  struct htt_rx_indication *rx)
{
	struct ath10k *ar = htt->ar;
	struct ieee80211_rx_status *rx_status = &htt->rx_status;
	struct htt_rx_indication_mpdu_range *mpdu_ranges;
	struct sk_buff_head amsdu;
	int num_mpdu_ranges;
	int fw_desc_len;
	u8 *fw_desc;
	int i, ret, mpdu_count = 0;

	lockdep_assert_held(&htt->rx_ring.lock);

	if (htt->rx_confused)
		return;

	fw_desc_len = __le16_to_cpu(rx->prefix.fw_rx_desc_bytes);
	fw_desc = (u8 *)&rx->fw_desc;

	num_mpdu_ranges = MS(__le32_to_cpu(rx->hdr.info1),
			     HTT_RX_INDICATION_INFO1_NUM_MPDU_RANGES);
	mpdu_ranges = htt_rx_ind_get_mpdu_ranges(rx);

	ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "htt rx ind: ",
			rx, sizeof(*rx) +
			(sizeof(struct htt_rx_indication_mpdu_range) *
				num_mpdu_ranges));

	for (i = 0; i < num_mpdu_ranges; i++)
		mpdu_count += mpdu_ranges[i].mpdu_count;

	while (mpdu_count--) {
		__skb_queue_head_init(&amsdu);
		ret = ath10k_htt_rx_amsdu_pop(htt, &fw_desc,
					      &fw_desc_len, &amsdu);
		if (ret < 0) {
			ath10k_warn(ar, "rx ring became corrupted: %d\n", ret);
			__skb_queue_purge(&amsdu);
			/* FIXME: It's probably a good idea to reboot the
			 * device instead of leaving it inoperable.
			 */
			htt->rx_confused = true;
			break;
		}

		ath10k_htt_rx_h_ppdu(ar, &amsdu, rx_status);
		ath10k_htt_rx_h_unchain(ar, &amsdu, ret > 0);
		ath10k_htt_rx_h_filter(ar, &amsdu, rx_status);
		ath10k_htt_rx_h_mpdu(ar, &amsdu, rx_status);
		ath10k_htt_rx_h_deliver(ar, &amsdu, rx_status);
	}

	tasklet_schedule(&htt->rx_replenish_task);
}

static void ath10k_htt_rx_frag_handler(struct ath10k_htt *htt,
				       struct htt_rx_fragment_indication *frag)
{
	struct ath10k *ar = htt->ar;
	struct ieee80211_rx_status *rx_status = &htt->rx_status;
	struct sk_buff_head amsdu;
	int ret;
	u8 *fw_desc;
	int fw_desc_len;

	fw_desc_len = __le16_to_cpu(frag->fw_rx_desc_bytes);
	fw_desc = (u8 *)frag->fw_msdu_rx_desc;

	__skb_queue_head_init(&amsdu);

	spin_lock_bh(&htt->rx_ring.lock);
	ret = ath10k_htt_rx_amsdu_pop(htt, &fw_desc, &fw_desc_len,
				      &amsdu);
	spin_unlock_bh(&htt->rx_ring.lock);

	tasklet_schedule(&htt->rx_replenish_task);

	ath10k_dbg(ar, ATH10K_DBG_HTT_DUMP, "htt rx frag ahead\n");

	if (ret) {
		ath10k_warn(ar, "failed to pop amsdu from httr rx ring for fragmented rx %d\n",
			    ret);
		__skb_queue_purge(&amsdu);
		return;
	}

	if (skb_queue_len(&amsdu) != 1) {
		ath10k_warn(ar, "failed to pop frag amsdu: too many msdus\n");
		__skb_queue_purge(&amsdu);
		return;
	}

	ath10k_htt_rx_h_ppdu(ar, &amsdu, rx_status);
	ath10k_htt_rx_h_filter(ar, &amsdu, rx_status);
	ath10k_htt_rx_h_mpdu(ar, &amsdu, rx_status);
	ath10k_htt_rx_h_deliver(ar, &amsdu, rx_status);

	if (fw_desc_len > 0) {
		ath10k_dbg(ar, ATH10K_DBG_HTT,
			   "expecting more fragmented rx in one indication %d\n",
			   fw_desc_len);
	}
}

static void ath10k_htt_rx_frm_tx_compl(struct ath10k *ar,
				       struct sk_buff *skb)
{
	struct ath10k_htt *htt = &ar->htt;
	struct htt_resp *resp = (struct htt_resp *)skb->data;
	struct htt_tx_done tx_done = {};
	int status = MS(resp->data_tx_completion.flags, HTT_DATA_TX_STATUS);
	__le16 msdu_id;
	int i;

	lockdep_assert_held(&htt->tx_lock);

	switch (status) {
	case HTT_DATA_TX_STATUS_NO_ACK:
		tx_done.no_ack = true;
		break;
	case HTT_DATA_TX_STATUS_OK:
		break;
	case HTT_DATA_TX_STATUS_DISCARD:
	case HTT_DATA_TX_STATUS_POSTPONE:
	case HTT_DATA_TX_STATUS_DOWNLOAD_FAIL:
		tx_done.discard = true;
		break;
	default:
		ath10k_warn(ar, "unhandled tx completion status %d\n", status);
		tx_done.discard = true;
		break;
	}

	ath10k_dbg(ar, ATH10K_DBG_HTT, "htt tx completion num_msdus %d\n",
		   resp->data_tx_completion.num_msdus);

	for (i = 0; i < resp->data_tx_completion.num_msdus; i++) {
		msdu_id = resp->data_tx_completion.msdus[i];
		tx_done.msdu_id = __le16_to_cpu(msdu_id);
		ath10k_txrx_tx_unref(htt, &tx_done);
	}
}

static void ath10k_htt_rx_addba(struct ath10k *ar, struct htt_resp *resp)
{
	struct htt_rx_addba *ev = &resp->rx_addba;
	struct ath10k_peer *peer;
	struct ath10k_vif *arvif;
	u16 info0, tid, peer_id;

	info0 = __le16_to_cpu(ev->info0);
	tid = MS(info0, HTT_RX_BA_INFO0_TID);
	peer_id = MS(info0, HTT_RX_BA_INFO0_PEER_ID);

	ath10k_dbg(ar, ATH10K_DBG_HTT,
		   "htt rx addba tid %hu peer_id %hu size %hhu\n",
		   tid, peer_id, ev->window_size);

	spin_lock_bh(&ar->data_lock);
	peer = ath10k_peer_find_by_id(ar, peer_id);
	if (!peer) {
		ath10k_warn(ar, "received addba event for invalid peer_id: %hu\n",
			    peer_id);
		spin_unlock_bh(&ar->data_lock);
		return;
	}

	arvif = ath10k_get_arvif(ar, peer->vdev_id);
	if (!arvif) {
		ath10k_warn(ar, "received addba event for invalid vdev_id: %u\n",
			    peer->vdev_id);
		spin_unlock_bh(&ar->data_lock);
		return;
	}

	ath10k_dbg(ar, ATH10K_DBG_HTT,
		   "htt rx start rx ba session sta %pM tid %hu size %hhu\n",
		   peer->addr, tid, ev->window_size);

	ieee80211_start_rx_ba_session_offl(arvif->vif, peer->addr, tid);
	spin_unlock_bh(&ar->data_lock);
}

static void ath10k_htt_rx_delba(struct ath10k *ar, struct htt_resp *resp)
{
	struct htt_rx_delba *ev = &resp->rx_delba;
	struct ath10k_peer *peer;
	struct ath10k_vif *arvif;
	u16 info0, tid, peer_id;

	info0 = __le16_to_cpu(ev->info0);
	tid = MS(info0, HTT_RX_BA_INFO0_TID);
	peer_id = MS(info0, HTT_RX_BA_INFO0_PEER_ID);

	ath10k_dbg(ar, ATH10K_DBG_HTT,
		   "htt rx delba tid %hu peer_id %hu\n",
		   tid, peer_id);

	spin_lock_bh(&ar->data_lock);
	peer = ath10k_peer_find_by_id(ar, peer_id);
	if (!peer) {
		ath10k_warn(ar, "received addba event for invalid peer_id: %hu\n",
			    peer_id);
		spin_unlock_bh(&ar->data_lock);
		return;
	}

	arvif = ath10k_get_arvif(ar, peer->vdev_id);
	if (!arvif) {
		ath10k_warn(ar, "received addba event for invalid vdev_id: %u\n",
			    peer->vdev_id);
		spin_unlock_bh(&ar->data_lock);
		return;
	}

	ath10k_dbg(ar, ATH10K_DBG_HTT,
		   "htt rx stop rx ba session sta %pM tid %hu\n",
		   peer->addr, tid);

	ieee80211_stop_rx_ba_session_offl(arvif->vif, peer->addr, tid);
	spin_unlock_bh(&ar->data_lock);
}

void ath10k_htt_t2h_msg_handler(struct ath10k *ar, struct sk_buff *skb)
{
	struct ath10k_htt *htt = &ar->htt;
	struct htt_resp *resp = (struct htt_resp *)skb->data;

	/* confirm alignment */
	if (!IS_ALIGNED((unsigned long)skb->data, 4))
		ath10k_warn(ar, "unaligned htt message, expect trouble\n");

	ath10k_dbg(ar, ATH10K_DBG_HTT, "htt rx, msg_type: 0x%0X\n",
		   resp->hdr.msg_type);
	switch (resp->hdr.msg_type) {
	case HTT_T2H_MSG_TYPE_VERSION_CONF: {
		htt->target_version_major = resp->ver_resp.major;
		htt->target_version_minor = resp->ver_resp.minor;
		complete(&htt->target_version_received);
		break;
	}
	case HTT_T2H_MSG_TYPE_RX_IND:
		spin_lock_bh(&htt->rx_ring.lock);
		__skb_queue_tail(&htt->rx_compl_q, skb);
		spin_unlock_bh(&htt->rx_ring.lock);
		tasklet_schedule(&htt->txrx_compl_task);
		return;
	case HTT_T2H_MSG_TYPE_PEER_MAP: {
		struct htt_peer_map_event ev = {
			.vdev_id = resp->peer_map.vdev_id,
			.peer_id = __le16_to_cpu(resp->peer_map.peer_id),
		};
		memcpy(ev.addr, resp->peer_map.addr, sizeof(ev.addr));
		ath10k_peer_map_event(htt, &ev);
		break;
	}
	case HTT_T2H_MSG_TYPE_PEER_UNMAP: {
		struct htt_peer_unmap_event ev = {
			.peer_id = __le16_to_cpu(resp->peer_unmap.peer_id),
		};
		ath10k_peer_unmap_event(htt, &ev);
		break;
	}
	case HTT_T2H_MSG_TYPE_MGMT_TX_COMPLETION: {
		struct htt_tx_done tx_done = {};
		int status = __le32_to_cpu(resp->mgmt_tx_completion.status);

		tx_done.msdu_id =
			__le32_to_cpu(resp->mgmt_tx_completion.desc_id);

		switch (status) {
		case HTT_MGMT_TX_STATUS_OK:
			break;
		case HTT_MGMT_TX_STATUS_RETRY:
			tx_done.no_ack = true;
			break;
		case HTT_MGMT_TX_STATUS_DROP:
			tx_done.discard = true;
			break;
		}

		spin_lock_bh(&htt->tx_lock);
		ath10k_txrx_tx_unref(htt, &tx_done);
		spin_unlock_bh(&htt->tx_lock);
		break;
	}
	case HTT_T2H_MSG_TYPE_TX_COMPL_IND:
		spin_lock_bh(&htt->tx_lock);
		__skb_queue_tail(&htt->tx_compl_q, skb);
		spin_unlock_bh(&htt->tx_lock);
		tasklet_schedule(&htt->txrx_compl_task);
		return;
	case HTT_T2H_MSG_TYPE_SEC_IND: {
		struct ath10k *ar = htt->ar;
		struct htt_security_indication *ev = &resp->security_indication;

		ath10k_dbg(ar, ATH10K_DBG_HTT,
			   "sec ind peer_id %d unicast %d type %d\n",
			  __le16_to_cpu(ev->peer_id),
			  !!(ev->flags & HTT_SECURITY_IS_UNICAST),
			  MS(ev->flags, HTT_SECURITY_TYPE));
		complete(&ar->install_key_done);
		break;
	}
	case HTT_T2H_MSG_TYPE_RX_FRAG_IND: {
		ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "htt event: ",
				skb->data, skb->len);
		ath10k_htt_rx_frag_handler(htt, &resp->rx_frag_ind);
		break;
	}
	case HTT_T2H_MSG_TYPE_TEST:
		/* FIX THIS */
		break;
	case HTT_T2H_MSG_TYPE_STATS_CONF:
		trace_ath10k_htt_stats(ar, skb->data, skb->len);
		break;
	case HTT_T2H_MSG_TYPE_TX_INSPECT_IND:
		/* Firmware can return tx frames if it's unable to fully
		 * process them and suspects host may be able to fix it. ath10k
		 * sends all tx frames as already inspected so this shouldn't
		 * happen unless fw has a bug.
		 */
		ath10k_warn(ar, "received an unexpected htt tx inspect event\n");
		break;
	case HTT_T2H_MSG_TYPE_RX_ADDBA:
		ath10k_htt_rx_addba(ar, resp);
		break;
	case HTT_T2H_MSG_TYPE_RX_DELBA:
		ath10k_htt_rx_delba(ar, resp);
		break;
	case HTT_T2H_MSG_TYPE_PKTLOG: {
		struct ath10k_pktlog_hdr *hdr =
			(struct ath10k_pktlog_hdr *)resp->pktlog_msg.payload;

		trace_ath10k_htt_pktlog(ar, resp->pktlog_msg.payload,
					sizeof(*hdr) +
					__le16_to_cpu(hdr->size));
		break;
	}
	case HTT_T2H_MSG_TYPE_RX_FLUSH: {
		/* Ignore this event because mac80211 takes care of Rx
		 * aggregation reordering.
		 */
		break;
	}
	default:
		ath10k_warn(ar, "htt event (%d) not handled\n",
			    resp->hdr.msg_type);
		ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "htt event: ",
				skb->data, skb->len);
		break;
	};

	/* Free the indication buffer */
	dev_kfree_skb_any(skb);
}

static void ath10k_htt_txrx_compl_task(unsigned long ptr)
{
	struct ath10k_htt *htt = (struct ath10k_htt *)ptr;
	struct htt_resp *resp;
	struct sk_buff *skb;

	spin_lock_bh(&htt->tx_lock);
	while ((skb = __skb_dequeue(&htt->tx_compl_q))) {
		ath10k_htt_rx_frm_tx_compl(htt->ar, skb);
		dev_kfree_skb_any(skb);
	}
	spin_unlock_bh(&htt->tx_lock);

	spin_lock_bh(&htt->rx_ring.lock);
	while ((skb = __skb_dequeue(&htt->rx_compl_q))) {
		resp = (struct htt_resp *)skb->data;
		ath10k_htt_rx_handler(htt, &resp->rx_ind);
		dev_kfree_skb_any(skb);
	}
	spin_unlock_bh(&htt->rx_ring.lock);
}