rxq->rx_ring = (struct hns3_desc *)rx_mz->addr;
rxq->rx_ring_phys_addr = rx_mz->iova;
- hns3_dbg(hw, "No.%u rx descriptors iova 0x%" PRIx64, q_info->idx,
- rxq->rx_ring_phys_addr);
-
return rxq;
}
txq->tx_ring = (struct hns3_desc *)tx_mz->addr;
txq->tx_ring_phys_addr = tx_mz->iova;
- hns3_dbg(hw, "No.%u tx descriptors iova 0x%" PRIx64, q_info->idx,
- txq->tx_ring_phys_addr);
-
/* Clear tx bd */
desc = txq->tx_ring;
for (i = 0; i < txq->nb_tx_desc; i++) {
/*
* For hns3 PF device, if the VLAN mode is HW_SHIFT_AND_DISCARD_MODE,
* the pvid_sw_discard_en in the queue struct should not be changed,
- * because PVID-related operations do not need to be processed by PMD
- * driver. For hns3 VF device, whether it needs to process PVID depends
+ * because PVID-related operations do not need to be processed by PMD.
+ * For hns3 VF device, whether it needs to process PVID depends
* on the configuration of PF kernel mode netdevice driver. And the
* related PF configuration is delivered through the mailbox and finally
- * reflectd in port_base_vlan_cfg.
+ * reflected in port_base_vlan_cfg.
*/
if (hns->is_vf || hw->vlan_mode == HNS3_SW_SHIFT_AND_DISCARD_MODE)
rxq->pvid_sw_discard_en = hw->port_base_vlan_cfg.state ==
return rte_mbuf_raw_alloc(rxq->mb_pool);
}
-static inline void
+static void
hns3_rx_ptp_timestamp_handle(struct hns3_rx_queue *rxq, struct rte_mbuf *mbuf,
- volatile struct hns3_desc *rxd)
+ uint64_t timestamp)
{
struct hns3_pf *pf = HNS3_DEV_PRIVATE_TO_PF(rxq->hns);
- uint64_t timestamp = rte_le_to_cpu_64(rxd->timestamp);
- mbuf->ol_flags |= RTE_MBUF_F_RX_IEEE1588_PTP | RTE_MBUF_F_RX_IEEE1588_TMST;
+ mbuf->ol_flags |= RTE_MBUF_F_RX_IEEE1588_PTP |
+ RTE_MBUF_F_RX_IEEE1588_TMST;
if (hns3_timestamp_rx_dynflag > 0) {
*RTE_MBUF_DYNFIELD(mbuf, hns3_timestamp_dynfield_offset,
rte_mbuf_timestamp_t *) = timestamp;
rxe->mbuf = nmb;
if (unlikely(bd_base_info & BIT(HNS3_RXD_TS_VLD_B)))
- hns3_rx_ptp_timestamp_handle(rxq, rxm, rxdp);
+ hns3_rx_ptp_timestamp_handle(rxq, rxm,
+ rte_le_to_cpu_64(rxdp->timestamp));
dma_addr = rte_mbuf_data_iova_default(nmb);
rxdp->addr = rte_cpu_to_le_64(dma_addr);
struct rte_mbuf *rxm;
struct rte_eth_dev *dev;
uint32_t bd_base_info;
+ uint64_t timestamp;
uint32_t l234_info;
uint32_t gro_size;
uint32_t ol_info;
rxm = rxe->mbuf;
rxe->mbuf = nmb;
+ if (unlikely(bd_base_info & BIT(HNS3_RXD_TS_VLD_B)))
+ timestamp = rte_le_to_cpu_64(rxdp->timestamp);
+
dma_addr = rte_cpu_to_le_64(rte_mbuf_data_iova_default(nmb));
rxdp->rx.bd_base_info = 0;
rxdp->addr = dma_addr;
}
if (unlikely(bd_base_info & BIT(HNS3_RXD_TS_VLD_B)))
- hns3_rx_ptp_timestamp_handle(rxq, first_seg, rxdp);
+ hns3_rx_ptp_timestamp_handle(rxq, first_seg, timestamp);
/*
* The last buffer of the received packet. packet len from
/*
* For hns3 PF device, if the VLAN mode is HW_SHIFT_AND_DISCARD_MODE,
* the pvid_sw_shift_en in the queue struct should not be changed,
- * because PVID-related operations do not need to be processed by PMD
- * driver. For hns3 VF device, whether it needs to process PVID depends
+ * because PVID-related operations do not need to be processed by PMD.
+ * For hns3 VF device, whether it needs to process PVID depends
* on the configuration of PF kernel mode netdev driver. And the
* related PF configuration is delivered through the mailbox and finally
- * reflectd in port_base_vlan_cfg.
+ * reflected in port_base_vlan_cfg.
*/
if (hns->is_vf || hw->vlan_mode == HNS3_SW_SHIFT_AND_DISCARD_MODE)
txq->pvid_sw_shift_en = hw->port_base_vlan_cfg.state ==
txq->min_tx_pkt_len = hw->min_tx_pkt_len;
txq->tso_mode = hw->tso_mode;
txq->udp_cksum_mode = hw->udp_cksum_mode;
+ txq->mbuf_fast_free_en = !!(dev->data->dev_conf.txmode.offloads &
+ RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE);
memset(&txq->basic_stats, 0, sizeof(struct hns3_tx_basic_stats));
memset(&txq->dfx_stats, 0, sizeof(struct hns3_tx_dfx_stats));
return 0;
}
-static void
+static int
hns3_tx_free_useless_buffer(struct hns3_tx_queue *txq)
{
uint16_t tx_next_clean = txq->next_to_clean;
- uint16_t tx_next_use = txq->next_to_use;
- uint16_t tx_bd_ready = txq->tx_bd_ready;
- uint16_t tx_bd_max = txq->nb_tx_desc;
- struct hns3_entry *tx_bak_pkt = &txq->sw_ring[tx_next_clean];
+ uint16_t tx_next_use = txq->next_to_use;
+ struct hns3_entry *tx_entry = &txq->sw_ring[tx_next_clean];
struct hns3_desc *desc = &txq->tx_ring[tx_next_clean];
- struct rte_mbuf *mbuf;
+ int i;
- while ((!(desc->tx.tp_fe_sc_vld_ra_ri &
- rte_cpu_to_le_16(BIT(HNS3_TXD_VLD_B)))) &&
- tx_next_use != tx_next_clean) {
- mbuf = tx_bak_pkt->mbuf;
- if (mbuf) {
- rte_pktmbuf_free_seg(mbuf);
- tx_bak_pkt->mbuf = NULL;
- }
+ if (tx_next_use >= tx_next_clean &&
+ tx_next_use < tx_next_clean + txq->tx_rs_thresh)
+ return -1;
- desc++;
- tx_bak_pkt++;
- tx_next_clean++;
- tx_bd_ready++;
-
- if (tx_next_clean >= tx_bd_max) {
- tx_next_clean = 0;
- desc = txq->tx_ring;
- tx_bak_pkt = txq->sw_ring;
- }
+ /*
+ * All mbufs can be released only when the VLD bits of all
+ * descriptors in a batch are cleared.
+ */
+ for (i = 0; i < txq->tx_rs_thresh; i++) {
+ if (desc[i].tx.tp_fe_sc_vld_ra_ri &
+ rte_le_to_cpu_16(BIT(HNS3_TXD_VLD_B)))
+ return -1;
}
- txq->next_to_clean = tx_next_clean;
- txq->tx_bd_ready = tx_bd_ready;
+ for (i = 0; i < txq->tx_rs_thresh; i++) {
+ rte_pktmbuf_free_seg(tx_entry[i].mbuf);
+ tx_entry[i].mbuf = NULL;
+ }
+
+ /* Update numbers of available descriptor due to buffer freed */
+ txq->tx_bd_ready += txq->tx_rs_thresh;
+ txq->next_to_clean += txq->tx_rs_thresh;
+ if (txq->next_to_clean >= txq->nb_tx_desc)
+ txq->next_to_clean = 0;
+
+ return 0;
+}
+
+static inline int
+hns3_tx_free_required_buffer(struct hns3_tx_queue *txq, uint16_t required_bds)
+{
+ while (required_bds > txq->tx_bd_ready) {
+ if (hns3_tx_free_useless_buffer(txq) != 0)
+ return -1;
+ }
+ return 0;
}
int
* in Tx direction based on hns3 network engine. So when the number of
* VLANs in the packets represented by rxm plus the number of VLAN
* offload by hardware such as PVID etc, exceeds two, the packets will
- * be discarded or the original VLAN of the packets will be overwitted
+ * be discarded or the original VLAN of the packets will be overwritten
* by hardware. When the PF PVID is enabled by calling the API function
* named rte_eth_dev_set_vlan_pvid or the VF PVID is enabled by the hns3
* PF kernel ether driver, the outer VLAN tag will always be the PVID.
/*
* The inner l2 length of mbuf is the sum of outer l4 length,
* tunneling header length and inner l2 length for a tunnel
- * packect. But in hns3 tx descriptor, the tunneling header
+ * packet. But in hns3 tx descriptor, the tunneling header
* length is contained in the field of outer L4 length.
* Therefore, driver need to calculate the outer L4 length and
* inner L2 length.
tmp_outer |= hns3_gen_field_val(HNS3_TXD_TUNTYPE_M,
HNS3_TXD_TUNTYPE_S, HNS3_TUN_NVGRE);
/*
- * For NVGRE tunnel packect, the outer L4 is empty. So only
+ * For NVGRE tunnel packet, the outer L4 is empty. So only
* fill the NVGRE header length to the outer L4 field.
*/
tmp_outer |= hns3_gen_field_val(HNS3_TXD_L4LEN_M,
* mbuf, but for hns3 descriptor, it is contained in the outer L4. So,
* there is a need that switching between them. To avoid multiple
* calculations, the length of the L2 header include the outer and
- * inner, will be filled during the parsing of tunnel packects.
+ * inner, will be filled during the parsing of tunnel packets.
*/
if (!(ol_flags & RTE_MBUF_F_TX_TUNNEL_MASK)) {
/*
if (ol_flags & RTE_MBUF_F_TX_OUTER_UDP_CKSUM) {
struct rte_udp_hdr *udp_hdr;
/*
- * If OUTER_UDP_CKSUM is support, HW can caclulate the pseudo
+ * If OUTER_UDP_CKSUM is support, HW can calculate the pseudo
* header for TSO packets
*/
if (ol_flags & RTE_MBUF_F_TX_TCP_SEG)
if (ol_flags & RTE_MBUF_F_TX_OUTER_UDP_CKSUM) {
struct rte_udp_hdr *udp_hdr;
/*
- * If OUTER_UDP_CKSUM is support, HW can caclulate the pseudo
+ * If OUTER_UDP_CKSUM is support, HW can calculate the pseudo
* header for TSO packets
*/
if (ol_flags & RTE_MBUF_F_TX_TCP_SEG)
tx_entry = &txq->sw_ring[txq->next_to_clean];
+ if (txq->mbuf_fast_free_en) {
+ rte_mempool_put_bulk(tx_entry->mbuf->pool,
+ (void **)tx_entry, txq->tx_rs_thresh);
+ for (i = 0; i < txq->tx_rs_thresh; i++)
+ tx_entry[i].mbuf = NULL;
+ goto update_field;
+ }
+
for (i = 0; i < txq->tx_rs_thresh; i++)
rte_prefetch0((tx_entry + i)->mbuf);
for (i = 0; i < txq->tx_rs_thresh; i++, tx_entry++) {
tx_entry->mbuf = NULL;
}
+update_field:
txq->next_to_clean = (tx_next_clean + 1) % txq->nb_tx_desc;
txq->tx_bd_ready += txq->tx_rs_thresh;
}
hns3_tx_setup_4bd(struct hns3_desc *txdp, struct rte_mbuf **pkts)
{
#define PER_LOOP_NUM 4
- const uint16_t bd_flag = BIT(HNS3_TXD_VLD_B) | BIT(HNS3_TXD_FE_B);
+ uint16_t bd_flag = BIT(HNS3_TXD_VLD_B) | BIT(HNS3_TXD_FE_B);
uint64_t dma_addr;
uint32_t i;
txdp->tx.paylen_fd_dop_ol4cs = 0;
txdp->tx.type_cs_vlan_tso_len = 0;
txdp->tx.ol_type_vlan_len_msec = 0;
+ if (unlikely((*pkts)->ol_flags & RTE_MBUF_F_TX_IEEE1588_TMST))
+ bd_flag |= BIT(HNS3_TXD_TSYN_B);
txdp->tx.tp_fe_sc_vld_ra_ri = rte_cpu_to_le_16(bd_flag);
}
}
static inline void
hns3_tx_setup_1bd(struct hns3_desc *txdp, struct rte_mbuf **pkts)
{
- const uint16_t bd_flag = BIT(HNS3_TXD_VLD_B) | BIT(HNS3_TXD_FE_B);
+ uint16_t bd_flag = BIT(HNS3_TXD_VLD_B) | BIT(HNS3_TXD_FE_B);
uint64_t dma_addr;
dma_addr = rte_mbuf_data_iova(*pkts);
txdp->tx.paylen_fd_dop_ol4cs = 0;
txdp->tx.type_cs_vlan_tso_len = 0;
txdp->tx.ol_type_vlan_len_msec = 0;
+ if (unlikely((*pkts)->ol_flags & RTE_MBUF_F_TX_IEEE1588_TMST))
+ bd_flag |= BIT(HNS3_TXD_TSYN_B);
txdp->tx.tp_fe_sc_vld_ra_ri = rte_cpu_to_le_16(bd_flag);
}
uint16_t nb_tx;
uint16_t i;
- /* free useless buffer */
- hns3_tx_free_useless_buffer(txq);
+ if (txq->tx_bd_ready < txq->tx_free_thresh)
+ (void)hns3_tx_free_useless_buffer(txq);
tx_next_use = txq->next_to_use;
tx_bd_max = txq->nb_tx_desc;
nb_buf = tx_pkt->nb_segs;
if (nb_buf > txq->tx_bd_ready) {
- txq->dfx_stats.queue_full_cnt++;
- if (nb_tx == 0)
- return 0;
-
- goto end_of_tx;
+ /* Try to release the required MBUF, but avoid releasing
+ * all MBUFs, otherwise, the MBUFs will be released for
+ * a long time and may cause jitter.
+ */
+ if (hns3_tx_free_required_buffer(txq, nb_buf) != 0) {
+ txq->dfx_stats.queue_full_cnt++;
+ goto end_of_tx;
+ }
}
/*
{
uint64_t offloads = dev->data->dev_conf.txmode.offloads;
- struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
- if (hns3_dev_get_support(hw, PTP))
- return false;
-
return (offloads == (offloads & RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE));
}
rx_mode.info, tx_mode.info);
}
-void hns3_set_rxtx_function(struct rte_eth_dev *eth_dev)
+static void
+hns3_eth_dev_fp_ops_config(const struct rte_eth_dev *dev)
+{
+ struct rte_eth_fp_ops *fpo = rte_eth_fp_ops;
+ uint16_t port_id = dev->data->port_id;
+
+ fpo[port_id].rx_pkt_burst = dev->rx_pkt_burst;
+ fpo[port_id].tx_pkt_burst = dev->tx_pkt_burst;
+ fpo[port_id].tx_pkt_prepare = dev->tx_pkt_prepare;
+ fpo[port_id].rx_descriptor_status = dev->rx_descriptor_status;
+ fpo[port_id].tx_descriptor_status = dev->tx_descriptor_status;
+}
+
+void
+hns3_set_rxtx_function(struct rte_eth_dev *eth_dev)
{
struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(eth_dev->data->dev_private);
struct hns3_adapter *hns = eth_dev->data->dev_private;
eth_dev->tx_pkt_burst = hns3_dummy_rxtx_burst;
eth_dev->tx_pkt_prepare = NULL;
}
+
+ hns3_eth_dev_fp_ops_config(eth_dev);
}
void
static int
hns3_tx_done_cleanup_full(struct hns3_tx_queue *txq, uint32_t free_cnt)
{
- uint16_t next_to_clean = txq->next_to_clean;
- uint16_t next_to_use = txq->next_to_use;
- uint16_t tx_bd_ready = txq->tx_bd_ready;
- struct hns3_entry *tx_pkt = &txq->sw_ring[next_to_clean];
- struct hns3_desc *desc = &txq->tx_ring[next_to_clean];
+ uint16_t round_free_cnt;
uint32_t idx;
if (free_cnt == 0 || free_cnt > txq->nb_tx_desc)
free_cnt = txq->nb_tx_desc;
- for (idx = 0; idx < free_cnt; idx++) {
- if (next_to_clean == next_to_use)
- break;
+ if (txq->tx_rs_thresh == 0)
+ return 0;
- if (desc->tx.tp_fe_sc_vld_ra_ri &
- rte_cpu_to_le_16(BIT(HNS3_TXD_VLD_B)))
+ round_free_cnt = roundup(free_cnt, txq->tx_rs_thresh);
+ for (idx = 0; idx < round_free_cnt; idx += txq->tx_rs_thresh) {
+ if (hns3_tx_free_useless_buffer(txq) != 0)
break;
-
- if (tx_pkt->mbuf != NULL) {
- rte_pktmbuf_free_seg(tx_pkt->mbuf);
- tx_pkt->mbuf = NULL;
- }
-
- next_to_clean++;
- tx_bd_ready++;
- tx_pkt++;
- desc++;
- if (next_to_clean == txq->nb_tx_desc) {
- tx_pkt = txq->sw_ring;
- desc = txq->tx_ring;
- next_to_clean = 0;
- }
- }
-
- if (idx > 0) {
- txq->next_to_clean = next_to_clean;
- txq->tx_bd_ready = tx_bd_ready;
}
- return (int)idx;
+ return RTE_MIN(idx, free_cnt);
}
int
{
dev->tx_pkt_burst = hns3_dummy_rxtx_burst;
dev->tx_pkt_prepare = NULL;
+ hns3_eth_dev_fp_ops_config(dev);
+
+ if (rte_eal_process_type() == RTE_PROC_SECONDARY)
+ return;
+
rte_wmb();
/* Disable tx datapath on secondary process. */
hns3_mp_req_stop_tx(dev);
dev->tx_pkt_burst = hns3_get_tx_function(dev, &prep);
dev->tx_pkt_prepare = prep;
+ hns3_eth_dev_fp_ops_config(dev);
+
+ if (rte_eal_process_type() == RTE_PROC_SECONDARY)
+ return;
+
hns3_mp_req_start_tx(dev);
}