path: root/drivers/net/sfc/sfc_rx.c

/*-
 *   BSD LICENSE
 *
 * Copyright (c) 2016-2017 Solarflare Communications Inc.
 * All rights reserved.
 *
 * This software was jointly developed between OKTET Labs (under contract
 * for Solarflare) and Solarflare Communications, Inc.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *
 * 1. Redistributions of source code must retain the above copyright notice,
 *    this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright notice,
 *    this list of conditions and the following disclaimer in the documentation
 *    and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
 * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
 * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

#include <rte_mempool.h>

#include "efx.h"

#include "sfc.h"
#include "sfc_debug.h"
#include "sfc_log.h"
#include "sfc_ev.h"
#include "sfc_rx.h"
#include "sfc_kvargs.h"
#include "sfc_tweak.h"

/*
 * Maximum number of Rx queue flush attempt in the case of failure or
 * flush timeout
 */
#define SFC_RX_QFLUSH_ATTEMPTS		(3)

/*
 * Time to wait between event queue polling attempts when waiting for Rx
 * queue flush done or failed events.
 */
#define SFC_RX_QFLUSH_POLL_WAIT_MS	(1)

/*
 * Maximum number of event queue polling attempts when waiting for Rx queue
 * flush done or failed events. It defines Rx queue flush attempt timeout
 * together with SFC_RX_QFLUSH_POLL_WAIT_MS.
 */
#define SFC_RX_QFLUSH_POLL_ATTEMPTS	(2000)

void
sfc_rx_qflush_done(struct sfc_rxq *rxq)
{
	rxq->state |= SFC_RXQ_FLUSHED;
	rxq->state &= ~SFC_RXQ_FLUSHING;
}

void
sfc_rx_qflush_failed(struct sfc_rxq *rxq)
{
	rxq->state |= SFC_RXQ_FLUSH_FAILED;
	rxq->state &= ~SFC_RXQ_FLUSHING;
}

static void
sfc_efx_rx_qrefill(struct sfc_efx_rxq *rxq)
{
	unsigned int free_space;
	unsigned int bulks;
	void *objs[SFC_RX_REFILL_BULK];
	efsys_dma_addr_t addr[RTE_DIM(objs)];
	unsigned int added = rxq->added;
	unsigned int id;
	unsigned int i;
	struct sfc_efx_rx_sw_desc *rxd;
	struct rte_mbuf *m;
	uint16_t port_id = rxq->dp.dpq.port_id;

	free_space = EFX_RXQ_LIMIT(rxq->ptr_mask + 1) -
		(added - rxq->completed);

	if (free_space < rxq->refill_threshold)
		return;

	bulks = free_space / RTE_DIM(objs);
	/* refill_threshold guarantees that bulks is positive */
	SFC_ASSERT(bulks > 0);

	id = added & rxq->ptr_mask;
	do {
		if (unlikely(rte_mempool_get_bulk(rxq->refill_mb_pool, objs,
						  RTE_DIM(objs)) < 0)) {
			/*
			 * It is hardly a safe way to increment counter
			 * from different contexts, but all PMDs do it.
			 */
			rxq->evq->sa->eth_dev->data->rx_mbuf_alloc_failed +=
				RTE_DIM(objs);
			/* Return if we have posted nothing yet */
			if (added == rxq->added)
				return;
			/* Push posted */
			break;
		}

		for (i = 0; i < RTE_DIM(objs);
		     ++i, id = (id + 1) & rxq->ptr_mask) {
			m = objs[i];

			rxd = &rxq->sw_desc[id];
			rxd->mbuf = m;

			SFC_ASSERT(rte_mbuf_refcnt_read(m) == 1);
			m->data_off = RTE_PKTMBUF_HEADROOM;
			SFC_ASSERT(m->next == NULL);
			SFC_ASSERT(m->nb_segs == 1);
			m->port = port_id;

			addr[i] = rte_pktmbuf_mtophys(m);
		}

		efx_rx_qpost(rxq->common, addr, rxq->buf_size,
			     RTE_DIM(objs), rxq->completed, added);
		added += RTE_DIM(objs);
	} while (--bulks > 0);

	SFC_ASSERT(added != rxq->added);
	rxq->added = added;
	efx_rx_qpush(rxq->common, added, &rxq->pushed);
}

static uint64_t
sfc_efx_rx_desc_flags_to_offload_flags(const unsigned int desc_flags)
{
	uint64_t mbuf_flags = 0;

	switch (desc_flags & (EFX_PKT_IPV4 | EFX_CKSUM_IPV4)) {
	case (EFX_PKT_IPV4 | EFX_CKSUM_IPV4):
		mbuf_flags |= PKT_RX_IP_CKSUM_GOOD;
		break;
	case EFX_PKT_IPV4:
		mbuf_flags |= PKT_RX_IP_CKSUM_BAD;
		break;
	default:
		RTE_BUILD_BUG_ON(PKT_RX_IP_CKSUM_UNKNOWN != 0);
		SFC_ASSERT((mbuf_flags & PKT_RX_IP_CKSUM_MASK) ==
			   PKT_RX_IP_CKSUM_UNKNOWN);
		break;
	}

	switch ((desc_flags &
		 (EFX_PKT_TCP | EFX_PKT_UDP | EFX_CKSUM_TCPUDP))) {
	case (EFX_PKT_TCP | EFX_CKSUM_TCPUDP):
	case (EFX_PKT_UDP | EFX_CKSUM_TCPUDP):
		mbuf_flags |= PKT_RX_L4_CKSUM_GOOD;
		break;
	case EFX_PKT_TCP:
	case EFX_PKT_UDP:
		mbuf_flags |= PKT_RX_L4_CKSUM_BAD;
		break;
	default:
		RTE_BUILD_BUG_ON(PKT_RX_L4_CKSUM_UNKNOWN != 0);
		SFC_ASSERT((mbuf_flags & PKT_RX_L4_CKSUM_MASK) ==
			   PKT_RX_L4_CKSUM_UNKNOWN);
		break;
	}

	return mbuf_flags;
}

static uint32_t
sfc_efx_rx_desc_flags_to_packet_type(const unsigned int desc_flags)
{
	return RTE_PTYPE_L2_ETHER |
		((desc_flags & EFX_PKT_IPV4) ?
			RTE_PTYPE_L3_IPV4_EXT_UNKNOWN : 0) |
		((desc_flags & EFX_PKT_IPV6) ?
			RTE_PTYPE_L3_IPV6_EXT_UNKNOWN : 0) |
		((desc_flags & EFX_PKT_TCP) ? RTE_PTYPE_L4_TCP : 0) |
		((desc_flags & EFX_PKT_UDP) ? RTE_PTYPE_L4_UDP : 0);
}

static const uint32_t *
sfc_efx_supported_ptypes_get(void)
{
	static const uint32_t ptypes[] = {
		RTE_PTYPE_L2_ETHER,
		RTE_PTYPE_L3_IPV4_EXT_UNKNOWN,
		RTE_PTYPE_L3_IPV6_EXT_UNKNOWN,
		RTE_PTYPE_L4_TCP,
		RTE_PTYPE_L4_UDP,
		RTE_PTYPE_UNKNOWN
	};

	return ptypes;
}

static void
sfc_efx_rx_set_rss_hash(struct sfc_efx_rxq *rxq, unsigned int flags,
			struct rte_mbuf *m)
{
#if EFSYS_OPT_RX_SCALE
	uint8_t *mbuf_data;


	if ((rxq->flags & SFC_EFX_RXQ_FLAG_RSS_HASH) == 0)
		return;

	mbuf_data = rte_pktmbuf_mtod(m, uint8_t *);

	if (flags & (EFX_PKT_IPV4 | EFX_PKT_IPV6)) {
		m->hash.rss = efx_pseudo_hdr_hash_get(rxq->common,
						      EFX_RX_HASHALG_TOEPLITZ,
						      mbuf_data);

		m->ol_flags |= PKT_RX_RSS_HASH;
	}
#endif
}

static uint16_t
sfc_efx_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
{
	struct sfc_dp_rxq *dp_rxq = rx_queue;
	struct sfc_efx_rxq *rxq = sfc_efx_rxq_by_dp_rxq(dp_rxq);
	unsigned int completed;
	unsigned int prefix_size = rxq->prefix_size;
	unsigned int done_pkts = 0;
	boolean_t discard_next = B_FALSE;
	struct rte_mbuf *scatter_pkt = NULL;

	if (unlikely((rxq->flags & SFC_EFX_RXQ_FLAG_RUNNING) == 0))
		return 0;

	sfc_ev_qpoll(rxq->evq);

	completed = rxq->completed;
	while (completed != rxq->pending && done_pkts < nb_pkts) {
		unsigned int id;
		struct sfc_efx_rx_sw_desc *rxd;
		struct rte_mbuf *m;
		unsigned int seg_len;
		unsigned int desc_flags;

		id = completed++ & rxq->ptr_mask;
		rxd = &rxq->sw_desc[id];
		m = rxd->mbuf;
		desc_flags = rxd->flags;

		if (discard_next)
			goto discard;

		if (desc_flags & (EFX_ADDR_MISMATCH | EFX_DISCARD))
			goto discard;

		if (desc_flags & EFX_PKT_PREFIX_LEN) {
			uint16_t tmp_size;
			int rc __rte_unused;

			rc = efx_pseudo_hdr_pkt_length_get(rxq->common,
				rte_pktmbuf_mtod(m, uint8_t *), &tmp_size);
			SFC_ASSERT(rc == 0);
			seg_len = tmp_size;
		} else {
			seg_len = rxd->size - prefix_size;
		}

		rte_pktmbuf_data_len(m) = seg_len;
		rte_pktmbuf_pkt_len(m) = seg_len;

		if (scatter_pkt != NULL) {
			if (rte_pktmbuf_chain(scatter_pkt, m) != 0) {
				rte_pktmbuf_free(scatter_pkt);
				goto discard;
			}
			/* The packet to deliver */
			m = scatter_pkt;
		}

		if (desc_flags & EFX_PKT_CONT) {
			/* The packet is scattered, more fragments to come */
			scatter_pkt = m;
			/* Further fragments have no prefix */
			prefix_size = 0;
			continue;
		}

		/* Scattered packet is done */
		scatter_pkt = NULL;
		/* The first fragment of the packet has prefix */
		prefix_size = rxq->prefix_size;

		m->ol_flags =
			sfc_efx_rx_desc_flags_to_offload_flags(desc_flags);
		m->packet_type =
			sfc_efx_rx_desc_flags_to_packet_type(desc_flags);

		/*
		 * Extract RSS hash from the packet prefix and
		 * set the corresponding field (if needed and possible)
		 */
		sfc_efx_rx_set_rss_hash(rxq, desc_flags, m);

		m->data_off += prefix_size;

		*rx_pkts++ = m;
		done_pkts++;
		continue;

discard:
		discard_next = ((desc_flags & EFX_PKT_CONT) != 0);
		rte_mempool_put(rxq->refill_mb_pool, m);
		rxd->mbuf = NULL;
	}

	/* pending is only moved when entire packet is received */
	SFC_ASSERT(scatter_pkt == NULL);

	rxq->completed = completed;

	sfc_efx_rx_qrefill(rxq);

	return done_pkts;
}

static sfc_dp_rx_qdesc_npending_t sfc_efx_rx_qdesc_npending;
static unsigned int
sfc_efx_rx_qdesc_npending(struct sfc_dp_rxq *dp_rxq)
{
	struct sfc_efx_rxq *rxq = sfc_efx_rxq_by_dp_rxq(dp_rxq);

	if ((rxq->flags & SFC_EFX_RXQ_FLAG_RUNNING) == 0)
		return 0;

	sfc_ev_qpoll(rxq->evq);

	return rxq->pending - rxq->completed;
}

struct sfc_rxq *
sfc_rxq_by_dp_rxq(const struct sfc_dp_rxq *dp_rxq)
{
	const struct sfc_dp_queue *dpq = &dp_rxq->dpq;
	struct rte_eth_dev *eth_dev;
	struct sfc_adapter *sa;
	struct sfc_rxq *rxq;

	SFC_ASSERT(rte_eth_dev_is_valid_port(dpq->port_id));
	eth_dev = &rte_eth_devices[dpq->port_id];

	sa = eth_dev->data->dev_private;

	SFC_ASSERT(dpq->queue_id < sa->rxq_count);
	rxq = sa->rxq_info[dpq->queue_id].rxq;

	SFC_ASSERT(rxq != NULL);
	return rxq;
}

static sfc_dp_rx_qcreate_t sfc_efx_rx_qcreate;
static int
sfc_efx_rx_qcreate(uint16_t port_id, uint16_t queue_id,
		   const struct rte_pci_addr *pci_addr, int socket_id,
		   const struct sfc_dp_rx_qcreate_info *info,
		   struct sfc_dp_rxq **dp_rxqp)
{
	struct sfc_efx_rxq *rxq;
	int rc;

	rc = ENOMEM;
	rxq = rte_zmalloc_socket("sfc-efx-rxq", sizeof(*rxq),
				 RTE_CACHE_LINE_SIZE, socket_id);
	if (rxq == NULL)
		goto fail_rxq_alloc;

	sfc_dp_queue_init(&rxq->dp.dpq, port_id, queue_id, pci_addr);

	rc = ENOMEM;
	rxq->sw_desc = rte_calloc_socket("sfc-efx-rxq-sw_desc",
					 info->rxq_entries,
					 sizeof(*rxq->sw_desc),
					 RTE_CACHE_LINE_SIZE, socket_id);
	if (rxq->sw_desc == NULL)
		goto fail_desc_alloc;

	/* efx datapath is bound to efx control path */
	rxq->evq = sfc_rxq_by_dp_rxq(&rxq->dp)->evq;
	if (info->flags & SFC_RXQ_FLAG_RSS_HASH)
		rxq->flags |= SFC_EFX_RXQ_FLAG_RSS_HASH;
	rxq->ptr_mask = info->rxq_entries - 1;
	rxq->batch_max = info->batch_max;
	rxq->prefix_size = info->prefix_size;
	rxq->refill_threshold = info->refill_threshold;
	rxq->buf_size = info->buf_size;
	rxq->refill_mb_pool = info->refill_mb_pool;

	*dp_rxqp = &rxq->dp;
	return 0;

fail_desc_alloc:
	rte_free(rxq);

fail_rxq_alloc:
	return rc;
}

static sfc_dp_rx_qdestroy_t sfc_efx_rx_qdestroy;
static void
sfc_efx_rx_qdestroy(struct sfc_dp_rxq *dp_rxq)
{
	struct sfc_efx_rxq *rxq = sfc_efx_rxq_by_dp_rxq(dp_rxq);

	rte_free(rxq->sw_desc);
	rte_free(rxq);
}

static sfc_dp_rx_qstart_t sfc_efx_rx_qstart;
static int
sfc_efx_rx_qstart(struct sfc_dp_rxq *dp_rxq,
		  __rte_unused unsigned int evq_read_ptr)
{
	/* libefx-based datapath is specific to libefx-based PMD */
	struct sfc_efx_rxq *rxq = sfc_efx_rxq_by_dp_rxq(dp_rxq);
	struct sfc_rxq *crxq = sfc_rxq_by_dp_rxq(dp_rxq);

	rxq->common = crxq->common;

	rxq->pending = rxq->completed = rxq->added = rxq->pushed = 0;

	sfc_efx_rx_qrefill(rxq);

	rxq->flags |= (SFC_EFX_RXQ_FLAG_STARTED | SFC_EFX_RXQ_FLAG_RUNNING);

	return 0;
}

static sfc_dp_rx_qstop_t sfc_efx_rx_qstop;
static void
sfc_efx_rx_qstop(struct sfc_dp_rxq *dp_rxq,
		 __rte_unused unsigned int *evq_read_ptr)
{
	struct sfc_efx_rxq *rxq = sfc_efx_rxq_by_dp_rxq(dp_rxq);

	rxq->flags &= ~SFC_EFX_RXQ_FLAG_RUNNING;

	/* libefx-based datapath is bound to libefx-based PMD and uses
	 * event queue structure directly. So, there is no necessity to
	 * return EvQ read pointer.
	 */
}

static sfc_dp_rx_qpurge_t sfc_efx_rx_qpurge;
static void
sfc_efx_rx_qpurge(struct sfc_dp_rxq *dp_rxq)
{
	struct sfc_efx_rxq *rxq = sfc_efx_rxq_by_dp_rxq(dp_rxq);
	unsigned int i;
	struct sfc_efx_rx_sw_desc *rxd;

	for (i = rxq->completed; i != rxq->added; ++i) {
		rxd = &rxq->sw_desc[i & rxq->ptr_mask];
		rte_mempool_put(rxq->refill_mb_pool, rxd->mbuf);
		rxd->mbuf = NULL;
		/* Packed stream relies on 0 in inactive SW desc.
		 * Rx queue stop is not performance critical, so
		 * there is no harm to do it always.
		 */
		rxd->flags = 0;
		rxd->size = 0;
	}

	rxq->flags &= ~SFC_EFX_RXQ_FLAG_STARTED;
}

struct sfc_dp_rx sfc_efx_rx = {
	.dp = {
		.name		= SFC_KVARG_DATAPATH_EFX,
		.type		= SFC_DP_RX,
		.hw_fw_caps	= 0,
	},
	.features		= SFC_DP_RX_FEAT_SCATTER,
	.qcreate		= sfc_efx_rx_qcreate,
	.qdestroy		= sfc_efx_rx_qdestroy,
	.qstart			= sfc_efx_rx_qstart,
	.qstop			= sfc_efx_rx_qstop,
	.qpurge			= sfc_efx_rx_qpurge,
	.supported_ptypes_get	= sfc_efx_supported_ptypes_get,
	.qdesc_npending		= sfc_efx_rx_qdesc_npending,
	.pkt_burst		= sfc_efx_recv_pkts,
};

unsigned int
sfc_rx_qdesc_npending(struct sfc_adapter *sa, unsigned int sw_index)
{
	struct sfc_rxq *rxq;

	SFC_ASSERT(sw_index < sa->rxq_count);
	rxq = sa->rxq_info[sw_index].rxq;

	if (rxq == NULL || (rxq->state & SFC_RXQ_STARTED) == 0)
		return 0;

	return sa->dp_rx->qdesc_npending(rxq->dp);
}

int
sfc_rx_qdesc_done(struct sfc_dp_rxq *dp_rxq, unsigned int offset)
{
	struct sfc_rxq *rxq = sfc_rxq_by_dp_rxq(dp_rxq);

	return offset < rxq->evq->sa->dp_rx->qdesc_npending(dp_rxq);
}

static void
sfc_rx_qflush(struct sfc_adapter *sa, unsigned int sw_index)
{
	struct sfc_rxq *rxq;
	unsigned int retry_count;
	unsigned int wait_count;
	int rc;

	rxq = sa->rxq_info[sw_index].rxq;
	SFC_ASSERT(rxq->state & SFC_RXQ_STARTED);

	/*
	 * Retry Rx queue flushing in the case of flush failed or
	 * timeout. In the worst case it can delay for 6 seconds.
	 */
	for (retry_count = 0;
	     ((rxq->state & SFC_RXQ_FLUSHED) == 0) &&
	     (retry_count < SFC_RX_QFLUSH_ATTEMPTS);
	     ++retry_count) {
		rc = efx_rx_qflush(rxq->common);
		if (rc != 0) {
			rxq->state |= (rc == EALREADY) ?
				SFC_RXQ_FLUSHED : SFC_RXQ_FLUSH_FAILED;
			break;
		}
		rxq->state &= ~SFC_RXQ_FLUSH_FAILED;
		rxq->state |= SFC_RXQ_FLUSHING;

		/*
		 * Wait for Rx queue flush done or failed event at least
		 * SFC_RX_QFLUSH_POLL_WAIT_MS milliseconds and not more
		 * than 2 seconds (SFC_RX_QFLUSH_POLL_WAIT_MS multiplied
		 * by SFC_RX_QFLUSH_POLL_ATTEMPTS).
		 */
		wait_count = 0;
		do {
			rte_delay_ms(SFC_RX_QFLUSH_POLL_WAIT_MS);
			sfc_ev_qpoll(rxq->evq);
		} while ((rxq->state & SFC_RXQ_FLUSHING) &&
			 (wait_count++ < SFC_RX_QFLUSH_POLL_ATTEMPTS));

		if (rxq->state & SFC_RXQ_FLUSHING)
			sfc_err(sa, "RxQ %u flush timed out", sw_index);

		if (rxq->state & SFC_RXQ_FLUSH_FAILED)
			sfc_err(sa, "RxQ %u flush failed", sw_index);

		if (rxq->state & SFC_RXQ_FLUSHED)
			sfc_info(sa, "RxQ %u flushed", sw_index);
	}

	sa->dp_rx->qpurge(rxq->dp);
}

static int
sfc_rx_default_rxq_set_filter(struct sfc_adapter *sa, struct sfc_rxq *rxq)
{
	boolean_t rss = (sa->rss_channels > 0) ? B_TRUE : B_FALSE;
	struct sfc_port *port = &sa->port;
	int rc;

	/*
	 * If promiscuous or all-multicast mode has been requested, setting
	 * filter for the default Rx queue might fail, in particular, while
	 * running over PCI function which is not a member of corresponding
	 * privilege groups; if this occurs, few iterations will be made to
	 * repeat this step without promiscuous and all-multicast flags set
	 */
retry:
	rc = efx_mac_filter_default_rxq_set(sa->nic, rxq->common, rss);
	if (rc == 0)
		return 0;
	else if (rc != EOPNOTSUPP)
		return rc;

	if (port->promisc) {
		sfc_warn(sa, "promiscuous mode has been requested, "
			     "but the HW rejects it");
		sfc_warn(sa, "promiscuous mode will be disabled");

		port->promisc = B_FALSE;
		rc = sfc_set_rx_mode(sa);
		if (rc != 0)
			return rc;

		goto retry;
	}

	if (port->allmulti) {
		sfc_warn(sa, "all-multicast mode has been requested, "
			     "but the HW rejects it");
		sfc_warn(sa, "all-multicast mode will be disabled");

		port->allmulti = B_FALSE;
		rc = sfc_set_rx_mode(sa);
		if (rc != 0)
			return rc;

		goto retry;
	}

	return rc;
}

int
sfc_rx_qstart(struct sfc_adapter *sa, unsigned int sw_index)
{
	struct sfc_port *port = &sa->port;
	struct sfc_rxq_info *rxq_info;
	struct sfc_rxq *rxq;
	struct sfc_evq *evq;
	int rc;

	sfc_log_init(sa, "sw_index=%u", sw_index);

	SFC_ASSERT(sw_index < sa->rxq_count);

	rxq_info = &sa->rxq_info[sw_index];
	rxq = rxq_info->rxq;
	SFC_ASSERT(rxq->state == SFC_RXQ_INITIALIZED);

	evq = rxq->evq;

	rc = sfc_ev_qstart(evq, sfc_evq_index_by_rxq_sw_index(sa, sw_index));
	if (rc != 0)
		goto fail_ev_qstart;

	rc = efx_rx_qcreate(sa->nic, rxq->hw_index, 0, rxq_info->type,
			    &rxq->mem, rxq_info->entries,
			    0 /* not used on EF10 */, evq->common,
			    &rxq->common);
	if (rc != 0)
		goto fail_rx_qcreate;

	efx_rx_qenable(rxq->common);

	rc = sa->dp_rx->qstart(rxq->dp, evq->read_ptr);
	if (rc != 0)
		goto fail_dp_qstart;

	rxq->state |= SFC_RXQ_STARTED;

	if ((sw_index == 0) && !port->isolated) {
		rc = sfc_rx_default_rxq_set_filter(sa, rxq);
		if (rc != 0)
			goto fail_mac_filter_default_rxq_set;
	}

	/* It seems to be used by DPDK for debug purposes only ('rte_ether') */
	sa->eth_dev->data->rx_queue_state[sw_index] =
		RTE_ETH_QUEUE_STATE_STARTED;

	return 0;

fail_mac_filter_default_rxq_set:
	sa->dp_rx->qstop(rxq->dp, &rxq->evq->read_ptr);

fail_dp_qstart:
	sfc_rx_qflush(sa, sw_index);

fail_rx_qcreate:
	sfc_ev_qstop(evq);

fail_ev_qstart:
	return rc;
}

void
sfc_rx_qstop(struct sfc_adapter *sa, unsigned int sw_index)
{
	struct sfc_rxq_info *rxq_info;
	struct sfc_rxq *rxq;

	sfc_log_init(sa, "sw_index=%u", sw_index);

	SFC_ASSERT(sw_index < sa->rxq_count);

	rxq_info = &sa->rxq_info[sw_index];
	rxq = rxq_info->rxq;

	if (rxq->state == SFC_RXQ_INITIALIZED)
		return;
	SFC_ASSERT(rxq->state & SFC_RXQ_STARTED);

	/* It seems to be used by DPDK for debug purposes only ('rte_ether') */
	sa->eth_dev->data->rx_queue_state[sw_index] =
		RTE_ETH_QUEUE_STATE_STOPPED;

	sa->dp_rx->qstop(rxq->dp, &rxq->evq->read_ptr);

	if (sw_index == 0)
		efx_mac_filter_default_rxq_clear(sa->nic);

	sfc_rx_qflush(sa, sw_index);

	rxq->state = SFC_RXQ_INITIALIZED;

	efx_rx_qdestroy(rxq->common);

	sfc_ev_qstop(rxq->evq);
}

static int
sfc_rx_qcheck_conf(struct sfc_adapter *sa, uint16_t nb_rx_desc,
		   const struct rte_eth_rxconf *rx_conf)
{
	const uint16_t rx_free_thresh_max = EFX_RXQ_LIMIT(nb_rx_desc);
	int rc = 0;

	if (rx_conf->rx_thresh.pthresh != 0 ||
	    rx_conf->rx_thresh.hthresh != 0 ||
	    rx_conf->rx_thresh.wthresh != 0) {
		sfc_err(sa,
			"RxQ prefetch/host/writeback thresholds are not supported");
		rc = EINVAL;
	}

	if (rx_conf->rx_free_thresh > rx_free_thresh_max) {
		sfc_err(sa,
			"RxQ free threshold too large: %u vs maximum %u",
			rx_conf->rx_free_thresh, rx_free_thresh_max);
		rc = EINVAL;
	}

	if (rx_conf->rx_drop_en == 0) {
		sfc_err(sa, "RxQ drop disable is not supported");
		rc = EINVAL;
	}

	return rc;
}

static unsigned int
sfc_rx_mbuf_data_alignment(struct rte_mempool *mb_pool)
{
	uint32_t data_off;
	uint32_t order;

	/* The mbuf object itself is always cache line aligned */
	order = rte_bsf32(RTE_CACHE_LINE_SIZE);

	/* Data offset from mbuf object start */
	data_off = sizeof(struct rte_mbuf) + rte_pktmbuf_priv_size(mb_pool) +
		RTE_PKTMBUF_HEADROOM;

	order = MIN(order, rte_bsf32(data_off));

	return 1u << (order - 1);
}

static uint16_t
sfc_rx_mb_pool_buf_size(struct sfc_adapter *sa, struct rte_mempool *mb_pool)
{
	const efx_nic_cfg_t *encp = efx_nic_cfg_get(sa->nic);
	const uint32_t nic_align_start = MAX(1, encp->enc_rx_buf_align_start);
	const uint32_t nic_align_end = MAX(1, encp->enc_rx_buf_align_end);
	uint16_t buf_size;
	unsigned int buf_aligned;
	unsigned int start_alignment;
	unsigned int end_padding_alignment;

	/* Below it is assumed that both alignments are power of 2 */
	SFC_ASSERT(rte_is_power_of_2(nic_align_start));
	SFC_ASSERT(rte_is_power_of_2(nic_align_end));

	/*
	 * mbuf is always cache line aligned, double-check
	 * that it meets rx buffer start alignment requirements.
	 */

	/* Start from mbuf pool data room size */
	buf_size = rte_pktmbuf_data_room_size(mb_pool);

	/* Remove headroom */
	if (buf_size <= RTE_PKTMBUF_HEADROOM) {
		sfc_err(sa,
			"RxQ mbuf pool %s object data room size %u is smaller than headroom %u",
			mb_pool->name, buf_size, RTE_PKTMBUF_HEADROOM);
		return 0;
	}
	buf_size -= RTE_PKTMBUF_HEADROOM;

	/* Calculate guaranteed data start alignment */
	buf_aligned = sfc_rx_mbuf_data_alignment(mb_pool);

	/* Reserve space for start alignment */
	if (buf_aligned < nic_align_start) {
		start_alignment = nic_align_start - buf_aligned;
		if (buf_size <= start_alignment) {
			sfc_err(sa,
				"RxQ mbuf pool %s object data room size %u is insufficient for headroom %u and buffer start alignment %u required by NIC",
				mb_pool->name,
				rte_pktmbuf_data_room_size(mb_pool),
				RTE_PKTMBUF_HEADROOM, start_alignment);
			return 0;
		}
		buf_aligned = nic_align_start;
		buf_size -= start_alignment;
	} else {
		start_alignment = 0;
	}

	/* Make sure that end padding does not write beyond the buffer */
	if (buf_aligned < nic_align_end) {
		/*
		 * Estimate space which can be lost. If guarnteed buffer
		 * size is odd, lost space is (nic_align_end - 1). More
		 * accurate formula is below.
		 */
		end_padding_alignment = nic_align_end -
			MIN(buf_aligned, 1u << (rte_bsf32(buf_size) - 1));
		if (buf_size <= end_padding_alignment) {
			sfc_err(sa,
				"RxQ mbuf pool %s object data room size %u is insufficient for headroom %u, buffer start alignment %u and end padding alignment %u required by NIC",
				mb_pool->name,
				rte_pktmbuf_data_room_size(mb_pool),
				RTE_PKTMBUF_HEADROOM, start_alignment,
				end_padding_alignment);
			return 0;
		}
		buf_size -= end_padding_alignment;
	} else {
		/*
		 * Start is aligned the same or better than end,
		 * just align length.
		 */
		buf_size = P2ALIGN(buf_size, nic_align_end);
	}

	return buf_size;
}

int
sfc_rx_qinit(struct sfc_adapter *sa, unsigned int sw_index,
	     uint16_t nb_rx_desc, unsigned int socket_id,
	     const struct rte_eth_rxconf *rx_conf,
	     struct rte_mempool *mb_pool)
{
	const efx_nic_cfg_t *encp = efx_nic_cfg_get(sa->nic);
	int rc;
	uint16_t buf_size;
	struct sfc_rxq_info *rxq_info;
	struct sfc_evq *evq;
	struct sfc_rxq *rxq;
	struct sfc_dp_rx_qcreate_info info;

	rc = sfc_rx_qcheck_conf(sa, nb_rx_desc, rx_conf);
	if (rc != 0)
		goto fail_bad_conf;

	buf_size = sfc_rx_mb_pool_buf_size(sa, mb_pool);
	if (buf_size == 0) {
		sfc_err(sa, "RxQ %u mbuf pool object size is too small",
			sw_index);
		rc = EINVAL;
		goto fail_bad_conf;
	}

	if ((buf_size < sa->port.pdu + encp->enc_rx_prefix_size) &&
	    !sa->eth_dev->data->dev_conf.rxmode.enable_scatter) {
		sfc_err(sa, "Rx scatter is disabled and RxQ %u mbuf pool "
			"object size is too small", sw_index);
		sfc_err(sa, "RxQ %u calculated Rx buffer size is %u vs "
			"PDU size %u plus Rx prefix %u bytes",
			sw_index, buf_size, (unsigned int)sa->port.pdu,
			encp->enc_rx_prefix_size);
		rc = EINVAL;
		goto fail_bad_conf;
	}

	SFC_ASSERT(sw_index < sa->rxq_count);
	rxq_info = &sa->rxq_info[sw_index];

	SFC_ASSERT(nb_rx_desc <= rxq_info->max_entries);
	rxq_info->entries = nb_rx_desc;
	rxq_info->type =
		sa->eth_dev->data->dev_conf.rxmode.enable_scatter ?
		EFX_RXQ_TYPE_SCATTER : EFX_RXQ_TYPE_DEFAULT;

	rc = sfc_ev_qinit(sa, SFC_EVQ_TYPE_RX, sw_index,
			  rxq_info->entries, socket_id, &evq);
	if (rc != 0)
		goto fail_ev_qinit;

	rc = ENOMEM;
	rxq = rte_zmalloc_socket("sfc-rxq", sizeof(*rxq), RTE_CACHE_LINE_SIZE,
				 socket_id);
	if (rxq == NULL)
		goto fail_rxq_alloc;

	rxq_info->rxq = rxq;

	rxq->evq = evq;
	rxq->hw_index = sw_index;
	rxq->refill_threshold =
		RTE_MAX(rx_conf->rx_free_thresh, SFC_RX_REFILL_BULK);
	rxq->refill_mb_pool = mb_pool;

	rc = sfc_dma_alloc(sa, "rxq", sw_index, EFX_RXQ_SIZE(rxq_info->entries),
			   socket_id, &rxq->mem);
	if (rc != 0)
		goto fail_dma_alloc;

	memset(&info, 0, sizeof(info));
	info.refill_mb_pool = rxq->refill_mb_pool;
	info.refill_threshold = rxq->refill_threshold;
	info.buf_size = buf_size;
	info.batch_max = encp->enc_rx_batch_max;
	info.prefix_size = encp->enc_rx_prefix_size;

#if EFSYS_OPT_RX_SCALE
	if (sa->hash_support == EFX_RX_HASH_AVAILABLE && sa->rss_channels > 0)
		info.flags |= SFC_RXQ_FLAG_RSS_HASH;
#endif

	info.rxq_entries = rxq_info->entries;
	info.rxq_hw_ring = rxq->mem.esm_base;
	info.evq_entries = rxq_info->entries;
	info.evq_hw_ring = evq->mem.esm_base;
	info.hw_index = rxq->hw_index;
	info.mem_bar = sa->mem_bar.esb_base;

	rc = sa->dp_rx->qcreate(sa->eth_dev->data->port_id, sw_index,
				&RTE_ETH_DEV_TO_PCI(sa->eth_dev)->addr,
				socket_id, &info, &rxq->dp);
	if (rc != 0)
		goto fail_dp_rx_qcreate;

	evq->dp_rxq = rxq->dp;

	rxq->state = SFC_RXQ_INITIALIZED;

	rxq_info->deferred_start = (rx_conf->rx_deferred_start != 0);

	return 0;

fail_dp_rx_qcreate:
	sfc_dma_free(sa, &rxq->mem);

fail_dma_alloc:
	rxq_info->rxq = NULL;
	rte_free(rxq);

fail_rxq_alloc:
	sfc_ev_qfini(evq);

fail_ev_qinit:
	rxq_info->entries = 0;

fail_bad_conf:
	sfc_log_init(sa, "failed %d", rc);
	return rc;
}

void
sfc_rx_qfini(struct sfc_adapter *sa, unsigned int sw_index)
{
	struct sfc_rxq_info *rxq_info;
	struct sfc_rxq *rxq;

	SFC_ASSERT(sw_index < sa->rxq_count);

	rxq_info = &sa->rxq_info[sw_index];

	rxq = rxq_info->rxq;
	SFC_ASSERT(rxq->state == SFC_RXQ_INITIALIZED);

	sa->dp_rx->qdestroy(rxq->dp);
	rxq->dp = NULL;

	rxq_info->rxq = NULL;
	rxq_info->entries = 0;

	sfc_dma_free(sa, &rxq->mem);

	sfc_ev_qfini(rxq->evq);
	rxq->evq = NULL;

	rte_free(rxq);
}

#if EFSYS_OPT_RX_SCALE
efx_rx_hash_type_t
sfc_rte_to_efx_hash_type(uint64_t rss_hf)
{
	efx_rx_hash_type_t efx_hash_types = 0;

	if ((rss_hf & (ETH_RSS_IPV4 | ETH_RSS_FRAG_IPV4 |
		       ETH_RSS_NONFRAG_IPV4_OTHER)) != 0)
		efx_hash_types |= EFX_RX_HASH_IPV4;

	if ((rss_hf & ETH_RSS_NONFRAG_IPV4_TCP) != 0)
		efx_hash_types |= EFX_RX_HASH_TCPIPV4;

	if ((rss_hf & (ETH_RSS_IPV6 | ETH_RSS_FRAG_IPV6 |
			ETH_RSS_NONFRAG_IPV6_OTHER | ETH_RSS_IPV6_EX)) != 0)
		efx_hash_types |= EFX_RX_HASH_IPV6;

	if ((rss_hf & (ETH_RSS_NONFRAG_IPV6_TCP | ETH_RSS_IPV6_TCP_EX)) != 0)
		efx_hash_types |= EFX_RX_HASH_TCPIPV6;

	return efx_hash_types;
}

uint64_t
sfc_efx_to_rte_hash_type(efx_rx_hash_type_t efx_hash_types)
{
	uint64_t rss_hf = 0;

	if ((efx_hash_types & EFX_RX_HASH_IPV4) != 0)
		rss_hf |= (ETH_RSS_IPV4 | ETH_RSS_FRAG_IPV4 |
			   ETH_RSS_NONFRAG_IPV4_OTHER);

	if ((efx_hash_types & EFX_RX_HASH_TCPIPV4) != 0)
		rss_hf |= ETH_RSS_NONFRAG_IPV4_TCP;

	if ((efx_hash_types & EFX_RX_HASH_IPV6) != 0)
		rss_hf |= (ETH_RSS_IPV6 | ETH_RSS_FRAG_IPV6 |
			   ETH_RSS_NONFRAG_IPV6_OTHER | ETH_RSS_IPV6_EX);

	if ((efx_hash_types & EFX_RX_HASH_TCPIPV6) != 0)
		rss_hf |= (ETH_RSS_NONFRAG_IPV6_TCP | ETH_RSS_IPV6_TCP_EX);

	return rss_hf;
}
#endif

static int
sfc_rx_rss_config(struct sfc_adapter *sa)
{
	int rc = 0;

#if EFSYS_OPT_RX_SCALE
	if (sa->rss_channels > 0) {
		rc = efx_rx_scale_mode_set(sa->nic, EFX_RX_HASHALG_TOEPLITZ,
					   sa->rss_hash_types, B_TRUE);
		if (rc != 0)
			goto finish;

		rc = efx_rx_scale_key_set(sa->nic, sa->rss_key,
					  sizeof(sa->rss_key));
		if (rc != 0)
			goto finish;

		rc = efx_rx_scale_tbl_set(sa->nic, sa->rss_tbl,
					  sizeof(sa->rss_tbl));
	}

finish:
#endif
	return rc;
}

int
sfc_rx_start(struct sfc_adapter *sa)
{
	unsigned int sw_index;
	int rc;

	sfc_log_init(sa, "rxq_count=%u", sa->rxq_count);

	rc = efx_rx_init(sa->nic);
	if (rc != 0)
		goto fail_rx_init;

	rc = sfc_rx_rss_config(sa);
	if (rc != 0)
		goto fail_rss_config;

	for (sw_index = 0; sw_index < sa->rxq_count; ++sw_index) {
		if ((!sa->rxq_info[sw_index].deferred_start ||
		     sa->rxq_info[sw_index].deferred_started)) {
			rc = sfc_rx_qstart(sa, sw_index);
			if (rc != 0)
				goto fail_rx_qstart;
		}
	}

	return 0;

fail_rx_qstart:
	while (sw_index-- > 0)
		sfc_rx_qstop(sa, sw_index);

fail_rss_config:
	efx_rx_fini(sa->nic);

fail_rx_init:
	sfc_log_init(sa, "failed %d", rc);
	return rc;
}

void
sfc_rx_stop(struct sfc_adapter *sa)
{
	unsigned int sw_index;

	sfc_log_init(sa, "rxq_count=%u", sa->rxq_count);

	sw_index = sa->rxq_count;
	while (sw_index-- > 0) {
		if (sa->rxq_info[sw_index].rxq != NULL)
			sfc_rx_qstop(sa, sw_index);
	}

	efx_rx_fini(sa->nic);
}

static int
sfc_rx_qinit_info(struct sfc_adapter *sa, unsigned int sw_index)
{
	struct sfc_rxq_info *rxq_info = &sa->rxq_info[sw_index];
	unsigned int max_entries;

	max_entries = EFX_RXQ_MAXNDESCS;
	SFC_ASSERT(rte_is_power_of_2(max_entries));

	rxq_info->max_entries = max_entries;

	return 0;
}

static int
sfc_rx_check_mode(struct sfc_adapter *sa, struct rte_eth_rxmode *rxmode)
{
	int rc = 0;

	switch (rxmode->mq_mode) {
	case ETH_MQ_RX_NONE:
		/* No special checks are required */
		break;
#if EFSYS_OPT_RX_SCALE
	case ETH_MQ_RX_RSS:
		if (sa->rss_support == EFX_RX_SCALE_UNAVAILABLE) {
			sfc_err(sa, "RSS is not available");
			rc = EINVAL;
		}
		break;
#endif
	default:
		sfc_err(sa, "Rx multi-queue mode %u not supported",
			rxmode->mq_mode);
		rc = EINVAL;
	}

	if (rxmode->header_split) {
		sfc_err(sa, "Header split on Rx not supported");
		rc = EINVAL;
	}

	if (rxmode->hw_vlan_filter) {
		sfc_err(sa, "HW VLAN filtering not supported");
		rc = EINVAL;
	}

	if (rxmode->hw_vlan_strip) {
		sfc_err(sa, "HW VLAN stripping not supported");
		rc = EINVAL;
	}

	if (rxmode->hw_vlan_extend) {
		sfc_err(sa,
			"Q-in-Q HW VLAN stripping not supported");
		rc = EINVAL;
	}

	if (!rxmode->hw_strip_crc) {
		sfc_warn(sa,
			 "FCS stripping control not supported - always stripped");
		rxmode->hw_strip_crc = 1;
	}

	if (rxmode->enable_scatter &&
	    (~sa->dp_rx->features & SFC_DP_RX_FEAT_SCATTER)) {
		sfc_err(sa, "Rx scatter not supported by %s datapath",
			sa->dp_rx->dp.name);
		rc = EINVAL;
	}

	if (rxmode->enable_lro) {
		sfc_err(sa, "LRO not supported");
		rc = EINVAL;
	}

	return rc;
}

/**
 * Destroy excess queues that are no longer needed after reconfiguration
 * or complete close.
 */
static void
sfc_rx_fini_queues(struct sfc_adapter *sa, unsigned int nb_rx_queues)
{
	int sw_index;

	SFC_ASSERT(nb_rx_queues <= sa->rxq_count);

	sw_index = sa->rxq_count;
	while (--sw_index >= (int)nb_rx_queues) {
		if (sa->rxq_info[sw_index].rxq != NULL)
			sfc_rx_qfini(sa, sw_index);
	}

	sa->rxq_count = nb_rx_queues;
}

/**
 * Initialize Rx subsystem.
 *
 * Called at device (re)configuration stage when number of receive queues is
 * specified together with other device level receive configuration.
 *
 * It should be used to allocate NUMA-unaware resources.
 */
int
sfc_rx_configure(struct sfc_adapter *sa)
{
	struct rte_eth_conf *dev_conf = &sa->eth_dev->data->dev_conf;
	const unsigned int nb_rx_queues = sa->eth_dev->data->nb_rx_queues;
	unsigned int sw_index;
	int rc;

	sfc_log_init(sa, "nb_rx_queues=%u (old %u)",
		     nb_rx_queues, sa->rxq_count);

	rc = sfc_rx_check_mode(sa, &dev_conf->rxmode);
	if (rc != 0)
		goto fail_check_mode;

	if (nb_rx_queues == sa->rxq_count)
		goto done;

	if (sa->rxq_info == NULL) {
		rc = ENOMEM;
		sa->rxq_info = rte_calloc_socket("sfc-rxqs", nb_rx_queues,
						 sizeof(sa->rxq_info[0]), 0,
						 sa->socket_id);
		if (sa->rxq_info == NULL)
			goto fail_rxqs_alloc;
	} else {
		struct sfc_rxq_info *new_rxq_info;

		if (nb_rx_queues < sa->rxq_count)
			sfc_rx_fini_queues(sa, nb_rx_queues);

		rc = ENOMEM;
		new_rxq_info =
			rte_realloc(sa->rxq_info,
				    nb_rx_queues * sizeof(sa->rxq_info[0]), 0);
		if (new_rxq_info == NULL && nb_rx_queues > 0)
			goto fail_rxqs_realloc;

		sa->rxq_info = new_rxq_info;
		if (nb_rx_queues > sa->rxq_count)
			memset(&sa->rxq_info[sa->rxq_count], 0,
			       (nb_rx_queues - sa->rxq_count) *
			       sizeof(sa->rxq_info[0]));
	}

	while (sa->rxq_count < nb_rx_queues) {
		rc = sfc_rx_qinit_info(sa, sa->rxq_count);
		if (rc != 0)
			goto fail_rx_qinit_info;

		sa->rxq_count++;
	}

#if EFSYS_OPT_RX_SCALE
	sa->rss_channels = (dev_conf->rxmode.mq_mode == ETH_MQ_RX_RSS) ?
			   MIN(sa->rxq_count, EFX_MAXRSS) : 0;

	if (sa->rss_channels > 0) {
		for (sw_index = 0; sw_index < EFX_RSS_TBL_SIZE; ++sw_index)
			sa->rss_tbl[sw_index] = sw_index % sa->rss_channels;
	}
#endif

done:
	return 0;

fail_rx_qinit_info:
fail_rxqs_realloc:
fail_rxqs_alloc:
	sfc_rx_close(sa);

fail_check_mode:
	sfc_log_init(sa, "failed %d", rc);
	return rc;
}

/**
 * Shutdown Rx subsystem.
 *
 * Called at device close stage, for example, before device shutdown.
 */
void
sfc_rx_close(struct sfc_adapter *sa)
{
	sfc_rx_fini_queues(sa, 0);

	sa->rss_channels = 0;

	rte_free(sa->rxq_info);
	sa->rxq_info = NULL;
}