+ rte_mbuf_refcnt_set(mb, 1);
+ mb->data_off = RTE_PKTMBUF_HEADROOM;
+
+ /* populate the descriptors */
+ dma_addr = rte_cpu_to_le_64(rte_mbuf_data_iova_default(mb));
+ TXGBE_RXD_HDRADDR(&rxdp[i], 0);
+ TXGBE_RXD_PKTADDR(&rxdp[i], dma_addr);
+ }
+
+ /* update state of internal queue structure */
+ rxq->rx_free_trigger = rxq->rx_free_trigger + rxq->rx_free_thresh;
+ if (rxq->rx_free_trigger >= rxq->nb_rx_desc)
+ rxq->rx_free_trigger = rxq->rx_free_thresh - 1;
+
+ /* no errors */
+ return 0;
+}
+
+static inline uint16_t
+txgbe_rx_fill_from_stage(struct txgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts)
+{
+ struct rte_mbuf **stage = &rxq->rx_stage[rxq->rx_next_avail];
+ int i;
+
+ /* how many packets are ready to return? */
+ nb_pkts = (uint16_t)RTE_MIN(nb_pkts, rxq->rx_nb_avail);
+
+ /* copy mbuf pointers to the application's packet list */
+ for (i = 0; i < nb_pkts; ++i)
+ rx_pkts[i] = stage[i];
+
+ /* update internal queue state */
+ rxq->rx_nb_avail = (uint16_t)(rxq->rx_nb_avail - nb_pkts);
+ rxq->rx_next_avail = (uint16_t)(rxq->rx_next_avail + nb_pkts);
+
+ return nb_pkts;
+}
+
+static inline uint16_t
+txgbe_rx_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts)
+{
+ struct txgbe_rx_queue *rxq = (struct txgbe_rx_queue *)rx_queue;
+ struct rte_eth_dev *dev = &rte_eth_devices[rxq->port_id];
+ uint16_t nb_rx = 0;
+
+ /* Any previously recv'd pkts will be returned from the Rx stage */
+ if (rxq->rx_nb_avail)
+ return txgbe_rx_fill_from_stage(rxq, rx_pkts, nb_pkts);
+
+ /* Scan the H/W ring for packets to receive */
+ nb_rx = (uint16_t)txgbe_rx_scan_hw_ring(rxq);
+
+ /* update internal queue state */
+ rxq->rx_next_avail = 0;
+ rxq->rx_nb_avail = nb_rx;
+ rxq->rx_tail = (uint16_t)(rxq->rx_tail + nb_rx);
+
+ /* if required, allocate new buffers to replenish descriptors */
+ if (rxq->rx_tail > rxq->rx_free_trigger) {
+ uint16_t cur_free_trigger = rxq->rx_free_trigger;
+
+ if (txgbe_rx_alloc_bufs(rxq, true) != 0) {
+ int i, j;
+
+ PMD_RX_LOG(DEBUG, "RX mbuf alloc failed port_id=%u "
+ "queue_id=%u", (uint16_t)rxq->port_id,
+ (uint16_t)rxq->queue_id);
+
+ dev->data->rx_mbuf_alloc_failed +=
+ rxq->rx_free_thresh;
+
+ /*
+ * Need to rewind any previous receives if we cannot
+ * allocate new buffers to replenish the old ones.
+ */
+ rxq->rx_nb_avail = 0;
+ rxq->rx_tail = (uint16_t)(rxq->rx_tail - nb_rx);
+ for (i = 0, j = rxq->rx_tail; i < nb_rx; ++i, ++j)
+ rxq->sw_ring[j].mbuf = rxq->rx_stage[i];
+
+ return 0;
+ }
+
+ /* update tail pointer */
+ rte_wmb();
+ txgbe_set32_relaxed(rxq->rdt_reg_addr, cur_free_trigger);
+ }
+
+ if (rxq->rx_tail >= rxq->nb_rx_desc)
+ rxq->rx_tail = 0;
+
+ /* received any packets this loop? */
+ if (rxq->rx_nb_avail)
+ return txgbe_rx_fill_from_stage(rxq, rx_pkts, nb_pkts);
+
+ return 0;
+}
+
+/* split requests into chunks of size RTE_PMD_TXGBE_RX_MAX_BURST */
+uint16_t
+txgbe_recv_pkts_bulk_alloc(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts)
+{
+ uint16_t nb_rx;
+
+ if (unlikely(nb_pkts == 0))
+ return 0;
+
+ if (likely(nb_pkts <= RTE_PMD_TXGBE_RX_MAX_BURST))
+ return txgbe_rx_recv_pkts(rx_queue, rx_pkts, nb_pkts);
+
+ /* request is relatively large, chunk it up */
+ nb_rx = 0;
+ while (nb_pkts) {
+ uint16_t ret, n;
+
+ n = (uint16_t)RTE_MIN(nb_pkts, RTE_PMD_TXGBE_RX_MAX_BURST);
+ ret = txgbe_rx_recv_pkts(rx_queue, &rx_pkts[nb_rx], n);
+ nb_rx = (uint16_t)(nb_rx + ret);
+ nb_pkts = (uint16_t)(nb_pkts - ret);
+ if (ret < n)
+ break;
+ }
+
+ return nb_rx;
+}
+
+uint16_t
+txgbe_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts)
+{
+ struct txgbe_rx_queue *rxq;
+ volatile struct txgbe_rx_desc *rx_ring;
+ volatile struct txgbe_rx_desc *rxdp;
+ struct txgbe_rx_entry *sw_ring;
+ struct txgbe_rx_entry *rxe;
+ struct rte_mbuf *rxm;
+ struct rte_mbuf *nmb;
+ struct txgbe_rx_desc rxd;
+ uint64_t dma_addr;
+ uint32_t staterr;
+ uint32_t pkt_info;
+ uint16_t pkt_len;
+ uint16_t rx_id;
+ uint16_t nb_rx;
+ uint16_t nb_hold;
+ uint64_t pkt_flags;
+
+ nb_rx = 0;
+ nb_hold = 0;
+ rxq = rx_queue;
+ rx_id = rxq->rx_tail;
+ rx_ring = rxq->rx_ring;
+ sw_ring = rxq->sw_ring;
+ struct rte_eth_dev *dev = &rte_eth_devices[rxq->port_id];
+ while (nb_rx < nb_pkts) {
+ /*
+ * The order of operations here is important as the DD status
+ * bit must not be read after any other descriptor fields.
+ * rx_ring and rxdp are pointing to volatile data so the order
+ * of accesses cannot be reordered by the compiler. If they were
+ * not volatile, they could be reordered which could lead to
+ * using invalid descriptor fields when read from rxd.
+ */
+ rxdp = &rx_ring[rx_id];
+ staterr = rxdp->qw1.lo.status;
+ if (!(staterr & rte_cpu_to_le_32(TXGBE_RXD_STAT_DD)))
+ break;
+ rxd = *rxdp;
+
+ /*
+ * End of packet.
+ *
+ * If the TXGBE_RXD_STAT_EOP flag is not set, the RX packet
+ * is likely to be invalid and to be dropped by the various
+ * validation checks performed by the network stack.
+ *
+ * Allocate a new mbuf to replenish the RX ring descriptor.
+ * If the allocation fails:
+ * - arrange for that RX descriptor to be the first one
+ * being parsed the next time the receive function is
+ * invoked [on the same queue].
+ *
+ * - Stop parsing the RX ring and return immediately.
+ *
+ * This policy do not drop the packet received in the RX
+ * descriptor for which the allocation of a new mbuf failed.
+ * Thus, it allows that packet to be later retrieved if
+ * mbuf have been freed in the mean time.
+ * As a side effect, holding RX descriptors instead of
+ * systematically giving them back to the NIC may lead to
+ * RX ring exhaustion situations.
+ * However, the NIC can gracefully prevent such situations
+ * to happen by sending specific "back-pressure" flow control
+ * frames to its peer(s).
+ */
+ PMD_RX_LOG(DEBUG, "port_id=%u queue_id=%u rx_id=%u "
+ "ext_err_stat=0x%08x pkt_len=%u",
+ (uint16_t)rxq->port_id, (uint16_t)rxq->queue_id,
+ (uint16_t)rx_id, (uint32_t)staterr,
+ (uint16_t)rte_le_to_cpu_16(rxd.qw1.hi.len));
+
+ nmb = rte_mbuf_raw_alloc(rxq->mb_pool);
+ if (nmb == NULL) {
+ PMD_RX_LOG(DEBUG, "RX mbuf alloc failed port_id=%u "
+ "queue_id=%u", (uint16_t)rxq->port_id,
+ (uint16_t)rxq->queue_id);
+ dev->data->rx_mbuf_alloc_failed++;
+ break;
+ }
+
+ nb_hold++;
+ rxe = &sw_ring[rx_id];
+ rx_id++;
+ if (rx_id == rxq->nb_rx_desc)
+ rx_id = 0;
+
+ /* Prefetch next mbuf while processing current one. */
+ rte_txgbe_prefetch(sw_ring[rx_id].mbuf);
+
+ /*
+ * When next RX descriptor is on a cache-line boundary,
+ * prefetch the next 4 RX descriptors and the next 8 pointers
+ * to mbufs.
+ */
+ if ((rx_id & 0x3) == 0) {
+ rte_txgbe_prefetch(&rx_ring[rx_id]);
+ rte_txgbe_prefetch(&sw_ring[rx_id]);
+ }
+
+ rxm = rxe->mbuf;
+ rxe->mbuf = nmb;
+ dma_addr = rte_cpu_to_le_64(rte_mbuf_data_iova_default(nmb));
+ TXGBE_RXD_HDRADDR(rxdp, 0);
+ TXGBE_RXD_PKTADDR(rxdp, dma_addr);
+
+ /*
+ * Initialize the returned mbuf.
+ * 1) setup generic mbuf fields:
+ * - number of segments,
+ * - next segment,
+ * - packet length,
+ * - RX port identifier.
+ * 2) integrate hardware offload data, if any:
+ * - RSS flag & hash,
+ * - IP checksum flag,
+ * - VLAN TCI, if any,
+ * - error flags.
+ */
+ pkt_len = (uint16_t)(rte_le_to_cpu_16(rxd.qw1.hi.len) -
+ rxq->crc_len);
+ rxm->data_off = RTE_PKTMBUF_HEADROOM;
+ rte_packet_prefetch((char *)rxm->buf_addr + rxm->data_off);
+ rxm->nb_segs = 1;
+ rxm->next = NULL;
+ rxm->pkt_len = pkt_len;
+ rxm->data_len = pkt_len;
+ rxm->port = rxq->port_id;
+
+ pkt_info = rte_le_to_cpu_32(rxd.qw0.dw0);
+ /* Only valid if PKT_RX_VLAN set in pkt_flags */
+ rxm->vlan_tci = rte_le_to_cpu_16(rxd.qw1.hi.tag);
+
+ pkt_flags = rx_desc_status_to_pkt_flags(staterr,
+ rxq->vlan_flags);
+ pkt_flags |= rx_desc_error_to_pkt_flags(staterr);
+ pkt_flags |= txgbe_rxd_pkt_info_to_pkt_flags(pkt_info);
+ rxm->ol_flags = pkt_flags;
+ rxm->packet_type = txgbe_rxd_pkt_info_to_pkt_type(pkt_info,
+ rxq->pkt_type_mask);
+
+ if (likely(pkt_flags & PKT_RX_RSS_HASH)) {
+ rxm->hash.rss = rte_le_to_cpu_32(rxd.qw0.dw1);
+ } else if (pkt_flags & PKT_RX_FDIR) {
+ rxm->hash.fdir.hash =
+ rte_le_to_cpu_16(rxd.qw0.hi.csum) &
+ TXGBE_ATR_HASH_MASK;
+ rxm->hash.fdir.id = rte_le_to_cpu_16(rxd.qw0.hi.ipid);
+ }
+ /*
+ * Store the mbuf address into the next entry of the array
+ * of returned packets.
+ */
+ rx_pkts[nb_rx++] = rxm;
+ }
+ rxq->rx_tail = rx_id;
+
+ /*
+ * If the number of free RX descriptors is greater than the RX free
+ * threshold of the queue, advance the Receive Descriptor Tail (RDT)
+ * register.
+ * Update the RDT with the value of the last processed RX descriptor
+ * minus 1, to guarantee that the RDT register is never equal to the
+ * RDH register, which creates a "full" ring situation from the
+ * hardware point of view...
+ */
+ nb_hold = (uint16_t)(nb_hold + rxq->nb_rx_hold);
+ if (nb_hold > rxq->rx_free_thresh) {
+ PMD_RX_LOG(DEBUG, "port_id=%u queue_id=%u rx_tail=%u "
+ "nb_hold=%u nb_rx=%u",
+ (uint16_t)rxq->port_id, (uint16_t)rxq->queue_id,
+ (uint16_t)rx_id, (uint16_t)nb_hold,
+ (uint16_t)nb_rx);
+ rx_id = (uint16_t)((rx_id == 0) ?
+ (rxq->nb_rx_desc - 1) : (rx_id - 1));
+ txgbe_set32(rxq->rdt_reg_addr, rx_id);
+ nb_hold = 0;
+ }
+ rxq->nb_rx_hold = nb_hold;
+ return nb_rx;
+}
+
+/**
+ * txgbe_fill_cluster_head_buf - fill the first mbuf of the returned packet
+ *
+ * Fill the following info in the HEAD buffer of the Rx cluster:
+ * - RX port identifier
+ * - hardware offload data, if any:
+ * - RSS flag & hash
+ * - IP checksum flag
+ * - VLAN TCI, if any
+ * - error flags
+ * @head HEAD of the packet cluster
+ * @desc HW descriptor to get data from
+ * @rxq Pointer to the Rx queue
+ */
+static inline void
+txgbe_fill_cluster_head_buf(struct rte_mbuf *head, struct txgbe_rx_desc *desc,
+ struct txgbe_rx_queue *rxq, uint32_t staterr)
+{
+ uint32_t pkt_info;
+ uint64_t pkt_flags;
+
+ head->port = rxq->port_id;
+
+ /* The vlan_tci field is only valid when PKT_RX_VLAN is
+ * set in the pkt_flags field.
+ */
+ head->vlan_tci = rte_le_to_cpu_16(desc->qw1.hi.tag);
+ pkt_info = rte_le_to_cpu_32(desc->qw0.dw0);
+ pkt_flags = rx_desc_status_to_pkt_flags(staterr, rxq->vlan_flags);
+ pkt_flags |= rx_desc_error_to_pkt_flags(staterr);
+ pkt_flags |= txgbe_rxd_pkt_info_to_pkt_flags(pkt_info);
+ head->ol_flags = pkt_flags;
+ head->packet_type = txgbe_rxd_pkt_info_to_pkt_type(pkt_info,
+ rxq->pkt_type_mask);
+
+ if (likely(pkt_flags & PKT_RX_RSS_HASH)) {
+ head->hash.rss = rte_le_to_cpu_32(desc->qw0.dw1);
+ } else if (pkt_flags & PKT_RX_FDIR) {
+ head->hash.fdir.hash = rte_le_to_cpu_16(desc->qw0.hi.csum)
+ & TXGBE_ATR_HASH_MASK;
+ head->hash.fdir.id = rte_le_to_cpu_16(desc->qw0.hi.ipid);
+ }
+}
+
+/**
+ * txgbe_recv_pkts_lro - receive handler for and LRO case.
+ *
+ * @rx_queue Rx queue handle
+ * @rx_pkts table of received packets
+ * @nb_pkts size of rx_pkts table
+ * @bulk_alloc if TRUE bulk allocation is used for a HW ring refilling
+ *
+ * Handles the Rx HW ring completions when RSC feature is configured. Uses an
+ * additional ring of txgbe_rsc_entry's that will hold the relevant RSC info.
+ *
+ * We use the same logic as in Linux and in FreeBSD txgbe drivers:
+ * 1) When non-EOP RSC completion arrives:
+ * a) Update the HEAD of the current RSC aggregation cluster with the new
+ * segment's data length.
+ * b) Set the "next" pointer of the current segment to point to the segment
+ * at the NEXTP index.
+ * c) Pass the HEAD of RSC aggregation cluster on to the next NEXTP entry
+ * in the sw_rsc_ring.
+ * 2) When EOP arrives we just update the cluster's total length and offload
+ * flags and deliver the cluster up to the upper layers. In our case - put it
+ * in the rx_pkts table.
+ *
+ * Returns the number of received packets/clusters (according to the "bulk
+ * receive" interface).
+ */
+static inline uint16_t
+txgbe_recv_pkts_lro(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts,
+ bool bulk_alloc)
+{
+ struct txgbe_rx_queue *rxq = rx_queue;
+ struct rte_eth_dev *dev = &rte_eth_devices[rxq->port_id];
+ volatile struct txgbe_rx_desc *rx_ring = rxq->rx_ring;
+ struct txgbe_rx_entry *sw_ring = rxq->sw_ring;
+ struct txgbe_scattered_rx_entry *sw_sc_ring = rxq->sw_sc_ring;
+ uint16_t rx_id = rxq->rx_tail;
+ uint16_t nb_rx = 0;
+ uint16_t nb_hold = rxq->nb_rx_hold;
+ uint16_t prev_id = rxq->rx_tail;
+
+ while (nb_rx < nb_pkts) {
+ bool eop;
+ struct txgbe_rx_entry *rxe;
+ struct txgbe_scattered_rx_entry *sc_entry;
+ struct txgbe_scattered_rx_entry *next_sc_entry = NULL;
+ struct txgbe_rx_entry *next_rxe = NULL;
+ struct rte_mbuf *first_seg;
+ struct rte_mbuf *rxm;
+ struct rte_mbuf *nmb = NULL;
+ struct txgbe_rx_desc rxd;
+ uint16_t data_len;
+ uint16_t next_id;
+ volatile struct txgbe_rx_desc *rxdp;
+ uint32_t staterr;
+
+next_desc:
+ /*
+ * The code in this whole file uses the volatile pointer to
+ * ensure the read ordering of the status and the rest of the
+ * descriptor fields (on the compiler level only!!!). This is so
+ * UGLY - why not to just use the compiler barrier instead? DPDK
+ * even has the rte_compiler_barrier() for that.
+ *
+ * But most importantly this is just wrong because this doesn't
+ * ensure memory ordering in a general case at all. For
+ * instance, DPDK is supposed to work on Power CPUs where
+ * compiler barrier may just not be enough!
+ *
+ * I tried to write only this function properly to have a
+ * starting point (as a part of an LRO/RSC series) but the
+ * compiler cursed at me when I tried to cast away the
+ * "volatile" from rx_ring (yes, it's volatile too!!!). So, I'm
+ * keeping it the way it is for now.
+ *
+ * The code in this file is broken in so many other places and
+ * will just not work on a big endian CPU anyway therefore the
+ * lines below will have to be revisited together with the rest
+ * of the txgbe PMD.
+ *
+ * TODO:
+ * - Get rid of "volatile" and let the compiler do its job.
+ * - Use the proper memory barrier (rte_rmb()) to ensure the
+ * memory ordering below.
+ */
+ rxdp = &rx_ring[rx_id];
+ staterr = rte_le_to_cpu_32(rxdp->qw1.lo.status);
+
+ if (!(staterr & TXGBE_RXD_STAT_DD))
+ break;
+
+ rxd = *rxdp;
+
+ PMD_RX_LOG(DEBUG, "port_id=%u queue_id=%u rx_id=%u "
+ "staterr=0x%x data_len=%u",
+ rxq->port_id, rxq->queue_id, rx_id, staterr,
+ rte_le_to_cpu_16(rxd.qw1.hi.len));
+
+ if (!bulk_alloc) {
+ nmb = rte_mbuf_raw_alloc(rxq->mb_pool);
+ if (nmb == NULL) {
+ PMD_RX_LOG(DEBUG, "RX mbuf alloc failed "
+ "port_id=%u queue_id=%u",
+ rxq->port_id, rxq->queue_id);
+
+ dev->data->rx_mbuf_alloc_failed++;
+ break;
+ }
+ } else if (nb_hold > rxq->rx_free_thresh) {
+ uint16_t next_rdt = rxq->rx_free_trigger;
+
+ if (!txgbe_rx_alloc_bufs(rxq, false)) {
+ rte_wmb();
+ txgbe_set32_relaxed(rxq->rdt_reg_addr,
+ next_rdt);
+ nb_hold -= rxq->rx_free_thresh;
+ } else {
+ PMD_RX_LOG(DEBUG, "RX bulk alloc failed "
+ "port_id=%u queue_id=%u",
+ rxq->port_id, rxq->queue_id);
+
+ dev->data->rx_mbuf_alloc_failed++;
+ break;
+ }
+ }
+
+ nb_hold++;
+ rxe = &sw_ring[rx_id];
+ eop = staterr & TXGBE_RXD_STAT_EOP;
+
+ next_id = rx_id + 1;
+ if (next_id == rxq->nb_rx_desc)
+ next_id = 0;
+
+ /* Prefetch next mbuf while processing current one. */
+ rte_txgbe_prefetch(sw_ring[next_id].mbuf);
+
+ /*
+ * When next RX descriptor is on a cache-line boundary,
+ * prefetch the next 4 RX descriptors and the next 4 pointers
+ * to mbufs.
+ */
+ if ((next_id & 0x3) == 0) {
+ rte_txgbe_prefetch(&rx_ring[next_id]);
+ rte_txgbe_prefetch(&sw_ring[next_id]);
+ }
+
+ rxm = rxe->mbuf;
+
+ if (!bulk_alloc) {
+ __le64 dma =
+ rte_cpu_to_le_64(rte_mbuf_data_iova_default(nmb));
+ /*
+ * Update RX descriptor with the physical address of the
+ * new data buffer of the new allocated mbuf.
+ */
+ rxe->mbuf = nmb;
+
+ rxm->data_off = RTE_PKTMBUF_HEADROOM;
+ TXGBE_RXD_HDRADDR(rxdp, 0);
+ TXGBE_RXD_PKTADDR(rxdp, dma);
+ } else {
+ rxe->mbuf = NULL;
+ }
+
+ /*
+ * Set data length & data buffer address of mbuf.
+ */
+ data_len = rte_le_to_cpu_16(rxd.qw1.hi.len);
+ rxm->data_len = data_len;
+
+ if (!eop) {
+ uint16_t nextp_id;
+ /*
+ * Get next descriptor index:
+ * - For RSC it's in the NEXTP field.
+ * - For a scattered packet - it's just a following
+ * descriptor.
+ */
+ if (TXGBE_RXD_RSCCNT(rxd.qw0.dw0))
+ nextp_id = TXGBE_RXD_NEXTP(staterr);
+ else
+ nextp_id = next_id;
+
+ next_sc_entry = &sw_sc_ring[nextp_id];
+ next_rxe = &sw_ring[nextp_id];
+ rte_txgbe_prefetch(next_rxe);
+ }
+
+ sc_entry = &sw_sc_ring[rx_id];
+ first_seg = sc_entry->fbuf;
+ sc_entry->fbuf = NULL;
+
+ /*
+ * If this is the first buffer of the received packet,
+ * set the pointer to the first mbuf of the packet and
+ * initialize its context.
+ * Otherwise, update the total length and the number of segments
+ * of the current scattered packet, and update the pointer to
+ * the last mbuf of the current packet.
+ */
+ if (first_seg == NULL) {
+ first_seg = rxm;
+ first_seg->pkt_len = data_len;
+ first_seg->nb_segs = 1;
+ } else {
+ first_seg->pkt_len += data_len;
+ first_seg->nb_segs++;
+ }
+
+ prev_id = rx_id;
+ rx_id = next_id;
+
+ /*
+ * If this is not the last buffer of the received packet, update
+ * the pointer to the first mbuf at the NEXTP entry in the
+ * sw_sc_ring and continue to parse the RX ring.
+ */
+ if (!eop && next_rxe) {
+ rxm->next = next_rxe->mbuf;
+ next_sc_entry->fbuf = first_seg;
+ goto next_desc;
+ }
+
+ /* Initialize the first mbuf of the returned packet */
+ txgbe_fill_cluster_head_buf(first_seg, &rxd, rxq, staterr);
+
+ /*
+ * Deal with the case, when HW CRC srip is disabled.
+ * That can't happen when LRO is enabled, but still could
+ * happen for scattered RX mode.
+ */
+ first_seg->pkt_len -= rxq->crc_len;
+ if (unlikely(rxm->data_len <= rxq->crc_len)) {
+ struct rte_mbuf *lp;
+
+ for (lp = first_seg; lp->next != rxm; lp = lp->next)
+ ;
+
+ first_seg->nb_segs--;
+ lp->data_len -= rxq->crc_len - rxm->data_len;
+ lp->next = NULL;
+ rte_pktmbuf_free_seg(rxm);
+ } else {
+ rxm->data_len -= rxq->crc_len;
+ }
+
+ /* Prefetch data of first segment, if configured to do so. */
+ rte_packet_prefetch((char *)first_seg->buf_addr +
+ first_seg->data_off);
+
+ /*
+ * Store the mbuf address into the next entry of the array
+ * of returned packets.
+ */
+ rx_pkts[nb_rx++] = first_seg;
+ }
+
+ /*
+ * Record index of the next RX descriptor to probe.
+ */
+ rxq->rx_tail = rx_id;
+
+ /*
+ * If the number of free RX descriptors is greater than the RX free
+ * threshold of the queue, advance the Receive Descriptor Tail (RDT)
+ * register.
+ * Update the RDT with the value of the last processed RX descriptor
+ * minus 1, to guarantee that the RDT register is never equal to the
+ * RDH register, which creates a "full" ring situation from the
+ * hardware point of view...
+ */
+ if (!bulk_alloc && nb_hold > rxq->rx_free_thresh) {
+ PMD_RX_LOG(DEBUG, "port_id=%u queue_id=%u rx_tail=%u "
+ "nb_hold=%u nb_rx=%u",
+ rxq->port_id, rxq->queue_id, rx_id, nb_hold, nb_rx);
+
+ rte_wmb();
+ txgbe_set32_relaxed(rxq->rdt_reg_addr, prev_id);
+ nb_hold = 0;
+ }
+
+ rxq->nb_rx_hold = nb_hold;
+ return nb_rx;
+}
+
+uint16_t
+txgbe_recv_pkts_lro_single_alloc(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts)
+{
+ return txgbe_recv_pkts_lro(rx_queue, rx_pkts, nb_pkts, false);
+}
+
+uint16_t
+txgbe_recv_pkts_lro_bulk_alloc(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts)
+{
+ return txgbe_recv_pkts_lro(rx_queue, rx_pkts, nb_pkts, true);
+}
+
+uint64_t
+txgbe_get_rx_queue_offloads(struct rte_eth_dev *dev __rte_unused)
+{
+ return DEV_RX_OFFLOAD_VLAN_STRIP;
+}
+
+uint64_t
+txgbe_get_rx_port_offloads(struct rte_eth_dev *dev)
+{
+ uint64_t offloads;
+ struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+ struct rte_eth_dev_sriov *sriov = &RTE_ETH_DEV_SRIOV(dev);
+
+ offloads = DEV_RX_OFFLOAD_IPV4_CKSUM |
+ DEV_RX_OFFLOAD_UDP_CKSUM |
+ DEV_RX_OFFLOAD_TCP_CKSUM |
+ DEV_RX_OFFLOAD_KEEP_CRC |
+ DEV_RX_OFFLOAD_JUMBO_FRAME |
+ DEV_RX_OFFLOAD_VLAN_FILTER |
+ DEV_RX_OFFLOAD_RSS_HASH |
+ DEV_RX_OFFLOAD_SCATTER;
+
+ if (!txgbe_is_vf(dev))
+ offloads |= (DEV_RX_OFFLOAD_VLAN_FILTER |
+ DEV_RX_OFFLOAD_QINQ_STRIP |
+ DEV_RX_OFFLOAD_VLAN_EXTEND);
+
+ /*
+ * RSC is only supported by PF devices in a non-SR-IOV
+ * mode.
+ */
+ if (hw->mac.type == txgbe_mac_raptor && !sriov->active)
+ offloads |= DEV_RX_OFFLOAD_TCP_LRO;
+
+ if (hw->mac.type == txgbe_mac_raptor)
+ offloads |= DEV_RX_OFFLOAD_MACSEC_STRIP;
+
+ offloads |= DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM;
+
+ return offloads;
+}
+
+static void __rte_cold
+txgbe_tx_queue_release_mbufs(struct txgbe_tx_queue *txq)
+{
+ unsigned int i;
+
+ if (txq->sw_ring != NULL) {
+ for (i = 0; i < txq->nb_tx_desc; i++) {
+ if (txq->sw_ring[i].mbuf != NULL) {
+ rte_pktmbuf_free_seg(txq->sw_ring[i].mbuf);
+ txq->sw_ring[i].mbuf = NULL;
+ }
+ }
+ }
+}
+
+static void __rte_cold
+txgbe_tx_free_swring(struct txgbe_tx_queue *txq)
+{
+ if (txq != NULL &&
+ txq->sw_ring != NULL)
+ rte_free(txq->sw_ring);
+}
+
+static void __rte_cold
+txgbe_tx_queue_release(struct txgbe_tx_queue *txq)
+{
+ if (txq != NULL && txq->ops != NULL) {
+ txq->ops->release_mbufs(txq);
+ txq->ops->free_swring(txq);
+ rte_free(txq);
+ }
+}
+
+void __rte_cold
+txgbe_dev_tx_queue_release(void *txq)
+{
+ txgbe_tx_queue_release(txq);
+}
+
+/* (Re)set dynamic txgbe_tx_queue fields to defaults */
+static void __rte_cold
+txgbe_reset_tx_queue(struct txgbe_tx_queue *txq)
+{
+ static const struct txgbe_tx_desc zeroed_desc = {0};
+ struct txgbe_tx_entry *txe = txq->sw_ring;
+ uint16_t prev, i;
+
+ /* Zero out HW ring memory */
+ for (i = 0; i < txq->nb_tx_desc; i++)
+ txq->tx_ring[i] = zeroed_desc;
+
+ /* Initialize SW ring entries */
+ prev = (uint16_t)(txq->nb_tx_desc - 1);
+ for (i = 0; i < txq->nb_tx_desc; i++) {
+ volatile struct txgbe_tx_desc *txd = &txq->tx_ring[i];
+
+ txd->dw3 = rte_cpu_to_le_32(TXGBE_TXD_DD);
+ txe[i].mbuf = NULL;
+ txe[i].last_id = i;
+ txe[prev].next_id = i;
+ prev = i;
+ }
+
+ txq->tx_next_dd = (uint16_t)(txq->tx_free_thresh - 1);
+ txq->tx_tail = 0;
+
+ /*
+ * Always allow 1 descriptor to be un-allocated to avoid
+ * a H/W race condition
+ */
+ txq->last_desc_cleaned = (uint16_t)(txq->nb_tx_desc - 1);
+ txq->nb_tx_free = (uint16_t)(txq->nb_tx_desc - 1);
+ txq->ctx_curr = 0;
+ memset((void *)&txq->ctx_cache, 0,
+ TXGBE_CTX_NUM * sizeof(struct txgbe_ctx_info));
+}
+
+static const struct txgbe_txq_ops def_txq_ops = {
+ .release_mbufs = txgbe_tx_queue_release_mbufs,
+ .free_swring = txgbe_tx_free_swring,
+ .reset = txgbe_reset_tx_queue,
+};
+
+/* Takes an ethdev and a queue and sets up the tx function to be used based on
+ * the queue parameters. Used in tx_queue_setup by primary process and then
+ * in dev_init by secondary process when attaching to an existing ethdev.
+ */
+void __rte_cold
+txgbe_set_tx_function(struct rte_eth_dev *dev, struct txgbe_tx_queue *txq)
+{
+ /* Use a simple Tx queue (no offloads, no multi segs) if possible */
+ if (txq->offloads == 0 &&
+ txq->tx_free_thresh >= RTE_PMD_TXGBE_TX_MAX_BURST) {
+ PMD_INIT_LOG(DEBUG, "Using simple tx code path");
+ dev->tx_pkt_burst = txgbe_xmit_pkts_simple;
+ dev->tx_pkt_prepare = NULL;
+ } else {
+ PMD_INIT_LOG(DEBUG, "Using full-featured tx code path");
+ PMD_INIT_LOG(DEBUG,
+ " - offloads = 0x%" PRIx64,
+ txq->offloads);
+ PMD_INIT_LOG(DEBUG,
+ " - tx_free_thresh = %lu [RTE_PMD_TXGBE_TX_MAX_BURST=%lu]",
+ (unsigned long)txq->tx_free_thresh,
+ (unsigned long)RTE_PMD_TXGBE_TX_MAX_BURST);
+ dev->tx_pkt_burst = txgbe_xmit_pkts;
+ dev->tx_pkt_prepare = txgbe_prep_pkts;
+ }
+}
+
+uint64_t
+txgbe_get_tx_queue_offloads(struct rte_eth_dev *dev)
+{
+ RTE_SET_USED(dev);
+
+ return 0;
+}
+
+uint64_t
+txgbe_get_tx_port_offloads(struct rte_eth_dev *dev)
+{
+ uint64_t tx_offload_capa;
+
+ tx_offload_capa =
+ DEV_TX_OFFLOAD_VLAN_INSERT |
+ DEV_TX_OFFLOAD_IPV4_CKSUM |
+ DEV_TX_OFFLOAD_UDP_CKSUM |
+ DEV_TX_OFFLOAD_TCP_CKSUM |
+ DEV_TX_OFFLOAD_SCTP_CKSUM |
+ DEV_TX_OFFLOAD_TCP_TSO |
+ DEV_TX_OFFLOAD_UDP_TSO |
+ DEV_TX_OFFLOAD_UDP_TNL_TSO |
+ DEV_TX_OFFLOAD_IP_TNL_TSO |
+ DEV_TX_OFFLOAD_VXLAN_TNL_TSO |
+ DEV_TX_OFFLOAD_GRE_TNL_TSO |
+ DEV_TX_OFFLOAD_IPIP_TNL_TSO |
+ DEV_TX_OFFLOAD_GENEVE_TNL_TSO |
+ DEV_TX_OFFLOAD_MULTI_SEGS;
+
+ if (!txgbe_is_vf(dev))
+ tx_offload_capa |= DEV_TX_OFFLOAD_QINQ_INSERT;
+
+ tx_offload_capa |= DEV_TX_OFFLOAD_MACSEC_INSERT;
+
+ tx_offload_capa |= DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM;
+
+ return tx_offload_capa;
+}
+
+int __rte_cold
+txgbe_dev_tx_queue_setup(struct rte_eth_dev *dev,
+ uint16_t queue_idx,
+ uint16_t nb_desc,
+ unsigned int socket_id,
+ const struct rte_eth_txconf *tx_conf)
+{
+ const struct rte_memzone *tz;
+ struct txgbe_tx_queue *txq;
+ struct txgbe_hw *hw;
+ uint16_t tx_free_thresh;
+ uint64_t offloads;
+
+ PMD_INIT_FUNC_TRACE();
+ hw = TXGBE_DEV_HW(dev);
+
+ offloads = tx_conf->offloads | dev->data->dev_conf.txmode.offloads;
+
+ /*
+ * Validate number of transmit descriptors.
+ * It must not exceed hardware maximum, and must be multiple
+ * of TXGBE_ALIGN.
+ */
+ if (nb_desc % TXGBE_TXD_ALIGN != 0 ||
+ nb_desc > TXGBE_RING_DESC_MAX ||
+ nb_desc < TXGBE_RING_DESC_MIN) {
+ return -EINVAL;
+ }
+
+ /*
+ * The TX descriptor ring will be cleaned after txq->tx_free_thresh
+ * descriptors are used or if the number of descriptors required
+ * to transmit a packet is greater than the number of free TX
+ * descriptors.
+ * One descriptor in the TX ring is used as a sentinel to avoid a
+ * H/W race condition, hence the maximum threshold constraints.
+ * When set to zero use default values.
+ */
+ tx_free_thresh = (uint16_t)((tx_conf->tx_free_thresh) ?
+ tx_conf->tx_free_thresh : DEFAULT_TX_FREE_THRESH);
+ if (tx_free_thresh >= (nb_desc - 3)) {
+ PMD_INIT_LOG(ERR, "tx_free_thresh must be less than the number of "
+ "TX descriptors minus 3. (tx_free_thresh=%u "
+ "port=%d queue=%d)",
+ (unsigned int)tx_free_thresh,
+ (int)dev->data->port_id, (int)queue_idx);
+ return -(EINVAL);
+ }
+
+ if ((nb_desc % tx_free_thresh) != 0) {
+ PMD_INIT_LOG(ERR, "tx_free_thresh must be a divisor of the "
+ "number of TX descriptors. (tx_free_thresh=%u "
+ "port=%d queue=%d)", (unsigned int)tx_free_thresh,
+ (int)dev->data->port_id, (int)queue_idx);
+ return -(EINVAL);
+ }
+
+ /* Free memory prior to re-allocation if needed... */
+ if (dev->data->tx_queues[queue_idx] != NULL) {
+ txgbe_tx_queue_release(dev->data->tx_queues[queue_idx]);
+ dev->data->tx_queues[queue_idx] = NULL;
+ }
+
+ /* First allocate the tx queue data structure */
+ txq = rte_zmalloc_socket("ethdev TX queue",
+ sizeof(struct txgbe_tx_queue),
+ RTE_CACHE_LINE_SIZE, socket_id);
+ if (txq == NULL)
+ return -ENOMEM;
+
+ /*
+ * Allocate TX ring hardware descriptors. A memzone large enough to
+ * handle the maximum ring size is allocated in order to allow for
+ * resizing in later calls to the queue setup function.
+ */
+ tz = rte_eth_dma_zone_reserve(dev, "tx_ring", queue_idx,
+ sizeof(struct txgbe_tx_desc) * TXGBE_RING_DESC_MAX,
+ TXGBE_ALIGN, socket_id);
+ if (tz == NULL) {
+ txgbe_tx_queue_release(txq);
+ return -ENOMEM;
+ }
+
+ txq->nb_tx_desc = nb_desc;
+ txq->tx_free_thresh = tx_free_thresh;
+ txq->pthresh = tx_conf->tx_thresh.pthresh;
+ txq->hthresh = tx_conf->tx_thresh.hthresh;
+ txq->wthresh = tx_conf->tx_thresh.wthresh;
+ txq->queue_id = queue_idx;
+ txq->reg_idx = (uint16_t)((RTE_ETH_DEV_SRIOV(dev).active == 0) ?
+ queue_idx : RTE_ETH_DEV_SRIOV(dev).def_pool_q_idx + queue_idx);
+ txq->port_id = dev->data->port_id;
+ txq->offloads = offloads;
+ txq->ops = &def_txq_ops;
+ txq->tx_deferred_start = tx_conf->tx_deferred_start;
+
+ /* Modification to set tail pointer for virtual function
+ * if vf is detected.
+ */
+ if (hw->mac.type == txgbe_mac_raptor_vf) {
+ txq->tdt_reg_addr = TXGBE_REG_ADDR(hw, TXGBE_TXWP(queue_idx));
+ txq->tdc_reg_addr = TXGBE_REG_ADDR(hw, TXGBE_TXCFG(queue_idx));
+ } else {
+ txq->tdt_reg_addr = TXGBE_REG_ADDR(hw,
+ TXGBE_TXWP(txq->reg_idx));
+ txq->tdc_reg_addr = TXGBE_REG_ADDR(hw,
+ TXGBE_TXCFG(txq->reg_idx));
+ }
+
+ txq->tx_ring_phys_addr = TMZ_PADDR(tz);
+ txq->tx_ring = (struct txgbe_tx_desc *)TMZ_VADDR(tz);
+
+ /* Allocate software ring */
+ txq->sw_ring = rte_zmalloc_socket("txq->sw_ring",
+ sizeof(struct txgbe_tx_entry) * nb_desc,
+ RTE_CACHE_LINE_SIZE, socket_id);
+ if (txq->sw_ring == NULL) {
+ txgbe_tx_queue_release(txq);
+ return -ENOMEM;
+ }
+ PMD_INIT_LOG(DEBUG, "sw_ring=%p hw_ring=%p dma_addr=0x%" PRIx64,
+ txq->sw_ring, txq->tx_ring, txq->tx_ring_phys_addr);
+
+ /* set up scalar TX function as appropriate */
+ txgbe_set_tx_function(dev, txq);
+
+ txq->ops->reset(txq);
+
+ dev->data->tx_queues[queue_idx] = txq;
+
+ return 0;
+}
+
+/**
+ * txgbe_free_sc_cluster - free the not-yet-completed scattered cluster
+ *
+ * The "next" pointer of the last segment of (not-yet-completed) RSC clusters
+ * in the sw_rsc_ring is not set to NULL but rather points to the next
+ * mbuf of this RSC aggregation (that has not been completed yet and still
+ * resides on the HW ring). So, instead of calling for rte_pktmbuf_free() we
+ * will just free first "nb_segs" segments of the cluster explicitly by calling
+ * an rte_pktmbuf_free_seg().
+ *
+ * @m scattered cluster head
+ */
+static void __rte_cold
+txgbe_free_sc_cluster(struct rte_mbuf *m)
+{
+ uint16_t i, nb_segs = m->nb_segs;
+ struct rte_mbuf *next_seg;
+
+ for (i = 0; i < nb_segs; i++) {
+ next_seg = m->next;
+ rte_pktmbuf_free_seg(m);
+ m = next_seg;
+ }
+}
+
+static void __rte_cold
+txgbe_rx_queue_release_mbufs(struct txgbe_rx_queue *rxq)
+{
+ unsigned int i;
+
+ if (rxq->sw_ring != NULL) {
+ for (i = 0; i < rxq->nb_rx_desc; i++) {
+ if (rxq->sw_ring[i].mbuf != NULL) {
+ rte_pktmbuf_free_seg(rxq->sw_ring[i].mbuf);
+ rxq->sw_ring[i].mbuf = NULL;
+ }
+ }
+ if (rxq->rx_nb_avail) {
+ for (i = 0; i < rxq->rx_nb_avail; ++i) {
+ struct rte_mbuf *mb;
+
+ mb = rxq->rx_stage[rxq->rx_next_avail + i];
+ rte_pktmbuf_free_seg(mb);
+ }
+ rxq->rx_nb_avail = 0;
+ }
+ }
+
+ if (rxq->sw_sc_ring)
+ for (i = 0; i < rxq->nb_rx_desc; i++)
+ if (rxq->sw_sc_ring[i].fbuf) {
+ txgbe_free_sc_cluster(rxq->sw_sc_ring[i].fbuf);
+ rxq->sw_sc_ring[i].fbuf = NULL;
+ }
+}
+
+static void __rte_cold
+txgbe_rx_queue_release(struct txgbe_rx_queue *rxq)
+{
+ if (rxq != NULL) {
+ txgbe_rx_queue_release_mbufs(rxq);
+ rte_free(rxq->sw_ring);
+ rte_free(rxq->sw_sc_ring);
+ rte_free(rxq);
+ }
+}
+
+void __rte_cold
+txgbe_dev_rx_queue_release(void *rxq)
+{
+ txgbe_rx_queue_release(rxq);
+}
+
+/*
+ * Check if Rx Burst Bulk Alloc function can be used.
+ * Return
+ * 0: the preconditions are satisfied and the bulk allocation function
+ * can be used.
+ * -EINVAL: the preconditions are NOT satisfied and the default Rx burst
+ * function must be used.
+ */
+static inline int __rte_cold
+check_rx_burst_bulk_alloc_preconditions(struct txgbe_rx_queue *rxq)
+{
+ int ret = 0;
+
+ /*
+ * Make sure the following pre-conditions are satisfied:
+ * rxq->rx_free_thresh >= RTE_PMD_TXGBE_RX_MAX_BURST
+ * rxq->rx_free_thresh < rxq->nb_rx_desc
+ * (rxq->nb_rx_desc % rxq->rx_free_thresh) == 0
+ * Scattered packets are not supported. This should be checked
+ * outside of this function.
+ */
+ if (!(rxq->rx_free_thresh >= RTE_PMD_TXGBE_RX_MAX_BURST)) {
+ PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions: "
+ "rxq->rx_free_thresh=%d, "
+ "RTE_PMD_TXGBE_RX_MAX_BURST=%d",
+ rxq->rx_free_thresh, RTE_PMD_TXGBE_RX_MAX_BURST);
+ ret = -EINVAL;
+ } else if (!(rxq->rx_free_thresh < rxq->nb_rx_desc)) {
+ PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions: "
+ "rxq->rx_free_thresh=%d, "
+ "rxq->nb_rx_desc=%d",
+ rxq->rx_free_thresh, rxq->nb_rx_desc);
+ ret = -EINVAL;
+ } else if (!((rxq->nb_rx_desc % rxq->rx_free_thresh) == 0)) {
+ PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions: "
+ "rxq->nb_rx_desc=%d, "
+ "rxq->rx_free_thresh=%d",
+ rxq->nb_rx_desc, rxq->rx_free_thresh);
+ ret = -EINVAL;
+ }
+
+ return ret;
+}
+
+/* Reset dynamic txgbe_rx_queue fields back to defaults */
+static void __rte_cold
+txgbe_reset_rx_queue(struct txgbe_adapter *adapter, struct txgbe_rx_queue *rxq)
+{
+ static const struct txgbe_rx_desc zeroed_desc = {
+ {{0}, {0} }, {{0}, {0} } };
+ unsigned int i;
+ uint16_t len = rxq->nb_rx_desc;
+
+ /*
+ * By default, the Rx queue setup function allocates enough memory for
+ * TXGBE_RING_DESC_MAX. The Rx Burst bulk allocation function requires
+ * extra memory at the end of the descriptor ring to be zero'd out.
+ */
+ if (adapter->rx_bulk_alloc_allowed)
+ /* zero out extra memory */
+ len += RTE_PMD_TXGBE_RX_MAX_BURST;
+
+ /*
+ * Zero out HW ring memory. Zero out extra memory at the end of
+ * the H/W ring so look-ahead logic in Rx Burst bulk alloc function
+ * reads extra memory as zeros.
+ */
+ for (i = 0; i < len; i++)
+ rxq->rx_ring[i] = zeroed_desc;
+
+ /*
+ * initialize extra software ring entries. Space for these extra
+ * entries is always allocated
+ */
+ memset(&rxq->fake_mbuf, 0x0, sizeof(rxq->fake_mbuf));
+ for (i = rxq->nb_rx_desc; i < len; ++i)
+ rxq->sw_ring[i].mbuf = &rxq->fake_mbuf;
+
+ rxq->rx_nb_avail = 0;
+ rxq->rx_next_avail = 0;
+ rxq->rx_free_trigger = (uint16_t)(rxq->rx_free_thresh - 1);
+ rxq->rx_tail = 0;
+ rxq->nb_rx_hold = 0;
+ rxq->pkt_first_seg = NULL;
+ rxq->pkt_last_seg = NULL;
+}
+
+int __rte_cold
+txgbe_dev_rx_queue_setup(struct rte_eth_dev *dev,
+ uint16_t queue_idx,
+ uint16_t nb_desc,
+ unsigned int socket_id,
+ const struct rte_eth_rxconf *rx_conf,
+ struct rte_mempool *mp)
+{
+ const struct rte_memzone *rz;
+ struct txgbe_rx_queue *rxq;
+ struct txgbe_hw *hw;
+ uint16_t len;
+ struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+ uint64_t offloads;
+
+ PMD_INIT_FUNC_TRACE();
+ hw = TXGBE_DEV_HW(dev);
+
+ offloads = rx_conf->offloads | dev->data->dev_conf.rxmode.offloads;