numam-dpdk/drivers/net/hns3/hns3_rxtx_vec_sve.c
Chengwen Feng f0c243a6cb net/hns3: support SVE Tx
This patch adds SVE vector instructions to optimize Tx burst process.

Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Signed-off-by: Huisong Li <lihuisong@huawei.com>
2020-10-16 19:48:19 +02:00

473 lines
15 KiB
C

/* SPDX-License-Identifier: BSD-3-Clause
* Copyright(c) 2020 Hisilicon Limited.
*/
#include <arm_sve.h>
#include <rte_io.h>
#include <rte_ethdev_driver.h>
#include "hns3_ethdev.h"
#include "hns3_rxtx.h"
#include "hns3_rxtx_vec.h"
#define PG16_128BIT svwhilelt_b16(0, 8)
#define PG16_256BIT svwhilelt_b16(0, 16)
#define PG32_256BIT svwhilelt_b32(0, 8)
#define PG64_64BIT svwhilelt_b64(0, 1)
#define PG64_128BIT svwhilelt_b64(0, 2)
#define PG64_256BIT svwhilelt_b64(0, 4)
#define PG64_ALLBIT svptrue_b64()
#define BD_SIZE 32
#define BD_FIELD_ADDR_OFFSET 0
#define BD_FIELD_L234_OFFSET 8
#define BD_FIELD_XLEN_OFFSET 12
#define BD_FIELD_RSS_OFFSET 16
#define BD_FIELD_OL_OFFSET 24
#define BD_FIELD_VALID_OFFSET 28
typedef struct {
uint32_t l234_info[HNS3_SVE_DEFAULT_DESCS_PER_LOOP];
uint32_t ol_info[HNS3_SVE_DEFAULT_DESCS_PER_LOOP];
uint32_t bd_base_info[HNS3_SVE_DEFAULT_DESCS_PER_LOOP];
} HNS3_SVE_KEY_FIELD_S;
static inline uint32_t
hns3_desc_parse_field_sve(struct hns3_rx_queue *rxq,
struct rte_mbuf **rx_pkts,
HNS3_SVE_KEY_FIELD_S *key,
uint32_t bd_vld_num)
{
uint32_t retcode = 0;
uint32_t cksum_err;
int ret, i;
for (i = 0; i < (int)bd_vld_num; i++) {
/* init rte_mbuf.rearm_data last 64-bit */
rx_pkts[i]->ol_flags = PKT_RX_RSS_HASH;
ret = hns3_handle_bdinfo(rxq, rx_pkts[i], key->bd_base_info[i],
key->l234_info[i], &cksum_err);
if (unlikely(ret)) {
retcode |= 1u << i;
continue;
}
rx_pkts[i]->packet_type = hns3_rx_calc_ptype(rxq,
key->l234_info[i], key->ol_info[i]);
if (likely(key->bd_base_info[i] & BIT(HNS3_RXD_L3L4P_B)))
hns3_rx_set_cksum_flag(rx_pkts[i],
rx_pkts[i]->packet_type, cksum_err);
}
return retcode;
}
static inline void
hns3_rx_prefetch_mbuf_sve(struct hns3_entry *sw_ring)
{
svuint64_t prf1st = svld1_u64(PG64_256BIT, (uint64_t *)&sw_ring[0]);
svuint64_t prf2st = svld1_u64(PG64_256BIT, (uint64_t *)&sw_ring[4]);
svprfd_gather_u64base(PG64_256BIT, prf1st, SV_PLDL1KEEP);
svprfd_gather_u64base(PG64_256BIT, prf2st, SV_PLDL1KEEP);
}
static inline uint16_t
hns3_recv_burst_vec_sve(struct hns3_rx_queue *__restrict rxq,
struct rte_mbuf **__restrict rx_pkts,
uint16_t nb_pkts,
uint64_t *bd_err_mask)
{
#define XLEN_ADJUST_LEN 32
#define RSS_ADJUST_LEN 16
#define GEN_VLD_U8_ZIP_INDEX svindex_s8(28, -4)
uint16_t rx_id = rxq->next_to_use;
struct hns3_entry *sw_ring = &rxq->sw_ring[rx_id];
struct hns3_desc *rxdp = &rxq->rx_ring[rx_id];
struct hns3_desc *rxdp2;
HNS3_SVE_KEY_FIELD_S key_field;
uint64_t bd_valid_num;
uint32_t parse_retcode;
uint16_t nb_rx = 0;
int pos, offset;
uint16_t xlen_adjust[XLEN_ADJUST_LEN] = {
0, 0xffff, 1, 0xffff, /* 1st mbuf: pkt_len and dat_len */
2, 0xffff, 3, 0xffff, /* 2st mbuf: pkt_len and dat_len */
4, 0xffff, 5, 0xffff, /* 3st mbuf: pkt_len and dat_len */
6, 0xffff, 7, 0xffff, /* 4st mbuf: pkt_len and dat_len */
8, 0xffff, 9, 0xffff, /* 5st mbuf: pkt_len and dat_len */
10, 0xffff, 11, 0xffff, /* 6st mbuf: pkt_len and dat_len */
12, 0xffff, 13, 0xffff, /* 7st mbuf: pkt_len and dat_len */
14, 0xffff, 15, 0xffff, /* 8st mbuf: pkt_len and dat_len */
};
uint32_t rss_adjust[RSS_ADJUST_LEN] = {
0, 0xffff, /* 1st mbuf: rss */
1, 0xffff, /* 2st mbuf: rss */
2, 0xffff, /* 3st mbuf: rss */
3, 0xffff, /* 4st mbuf: rss */
4, 0xffff, /* 5st mbuf: rss */
5, 0xffff, /* 6st mbuf: rss */
6, 0xffff, /* 7st mbuf: rss */
7, 0xffff, /* 8st mbuf: rss */
};
svbool_t pg32 = svwhilelt_b32(0, HNS3_SVE_DEFAULT_DESCS_PER_LOOP);
svuint16_t xlen_tbl1 = svld1_u16(PG16_256BIT, xlen_adjust);
svuint16_t xlen_tbl2 = svld1_u16(PG16_256BIT, &xlen_adjust[16]);
svuint32_t rss_tbl1 = svld1_u32(PG32_256BIT, rss_adjust);
svuint32_t rss_tbl2 = svld1_u32(PG32_256BIT, &rss_adjust[8]);
for (pos = 0; pos < nb_pkts; pos += HNS3_SVE_DEFAULT_DESCS_PER_LOOP,
rxdp += HNS3_SVE_DEFAULT_DESCS_PER_LOOP) {
svuint64_t vld_clz, mbp1st, mbp2st, mbuf_init;
svuint64_t xlen1st, xlen2st, rss1st, rss2st;
svuint32_t l234, ol, vld, vld2, xlen, rss;
svuint8_t vld_u8;
/* calc how many bd valid: part 1 */
vld = svld1_gather_u32offset_u32(pg32, (uint32_t *)rxdp,
svindex_u32(BD_FIELD_VALID_OFFSET, BD_SIZE));
vld2 = svlsl_n_u32_z(pg32, vld,
HNS3_UINT32_BIT - 1 - HNS3_RXD_VLD_B);
vld2 = svreinterpret_u32_s32(svasr_n_s32_z(pg32,
svreinterpret_s32_u32(vld2), HNS3_UINT32_BIT - 1));
/* load 4 mbuf pointer */
mbp1st = svld1_u64(PG64_256BIT, (uint64_t *)&sw_ring[pos]);
/* calc how many bd valid: part 2 */
vld_u8 = svtbl_u8(svreinterpret_u8_u32(vld2),
svreinterpret_u8_s8(GEN_VLD_U8_ZIP_INDEX));
vld_clz = svnot_u64_z(PG64_64BIT, svreinterpret_u64_u8(vld_u8));
vld_clz = svclz_u64_z(PG64_64BIT, vld_clz);
svst1_u64(PG64_64BIT, &bd_valid_num, vld_clz);
bd_valid_num /= HNS3_UINT8_BIT;
/* load 4 more mbuf pointer */
mbp2st = svld1_u64(PG64_256BIT, (uint64_t *)&sw_ring[pos + 4]);
/* use offset to control below data load oper ordering */
offset = rxq->offset_table[bd_valid_num];
rxdp2 = rxdp + offset;
/* store 4 mbuf pointer into rx_pkts */
svst1_u64(PG64_256BIT, (uint64_t *)&rx_pkts[pos], mbp1st);
/* load key field to vector reg */
l234 = svld1_gather_u32offset_u32(pg32, (uint32_t *)rxdp2,
svindex_u32(BD_FIELD_L234_OFFSET, BD_SIZE));
ol = svld1_gather_u32offset_u32(pg32, (uint32_t *)rxdp2,
svindex_u32(BD_FIELD_OL_OFFSET, BD_SIZE));
/* store 4 mbuf pointer into rx_pkts again */
svst1_u64(PG64_256BIT, (uint64_t *)&rx_pkts[pos + 4], mbp2st);
/* load datalen, pktlen and rss_hash */
xlen = svld1_gather_u32offset_u32(pg32, (uint32_t *)rxdp2,
svindex_u32(BD_FIELD_XLEN_OFFSET, BD_SIZE));
rss = svld1_gather_u32offset_u32(pg32, (uint32_t *)rxdp2,
svindex_u32(BD_FIELD_RSS_OFFSET, BD_SIZE));
/* store key field to stash buffer */
svst1_u32(pg32, (uint32_t *)key_field.l234_info, l234);
svst1_u32(pg32, (uint32_t *)key_field.bd_base_info, vld);
svst1_u32(pg32, (uint32_t *)key_field.ol_info, ol);
/* sub crc_len for pkt_len and data_len */
xlen = svreinterpret_u32_u16(svsub_n_u16_z(PG16_256BIT,
svreinterpret_u16_u32(xlen), rxq->crc_len));
/* init mbuf_initializer */
mbuf_init = svdup_n_u64(rxq->mbuf_initializer);
/* extract datalen, pktlen and rss from xlen and rss */
xlen1st = svreinterpret_u64_u16(
svtbl_u16(svreinterpret_u16_u32(xlen), xlen_tbl1));
xlen2st = svreinterpret_u64_u16(
svtbl_u16(svreinterpret_u16_u32(xlen), xlen_tbl2));
rss1st = svreinterpret_u64_u32(
svtbl_u32(svreinterpret_u32_u32(rss), rss_tbl1));
rss2st = svreinterpret_u64_u32(
svtbl_u32(svreinterpret_u32_u32(rss), rss_tbl2));
/* save mbuf_initializer */
svst1_scatter_u64base_offset_u64(PG64_256BIT, mbp1st,
offsetof(struct rte_mbuf, rearm_data), mbuf_init);
svst1_scatter_u64base_offset_u64(PG64_256BIT, mbp2st,
offsetof(struct rte_mbuf, rearm_data), mbuf_init);
/* save datalen and pktlen and rss */
svst1_scatter_u64base_offset_u64(PG64_256BIT, mbp1st,
offsetof(struct rte_mbuf, pkt_len), xlen1st);
svst1_scatter_u64base_offset_u64(PG64_256BIT, mbp1st,
offsetof(struct rte_mbuf, hash.rss), rss1st);
svst1_scatter_u64base_offset_u64(PG64_256BIT, mbp2st,
offsetof(struct rte_mbuf, pkt_len), xlen2st);
svst1_scatter_u64base_offset_u64(PG64_256BIT, mbp2st,
offsetof(struct rte_mbuf, hash.rss), rss2st);
rte_prefetch_non_temporal(rxdp +
HNS3_SVE_DEFAULT_DESCS_PER_LOOP);
parse_retcode = hns3_desc_parse_field_sve(rxq, &rx_pkts[pos],
&key_field, bd_valid_num);
if (unlikely(parse_retcode))
(*bd_err_mask) |= ((uint64_t)parse_retcode) << pos;
hns3_rx_prefetch_mbuf_sve(&sw_ring[pos +
HNS3_SVE_DEFAULT_DESCS_PER_LOOP]);
nb_rx += bd_valid_num;
if (unlikely(bd_valid_num < HNS3_SVE_DEFAULT_DESCS_PER_LOOP))
break;
}
rxq->rx_rearm_nb += nb_rx;
rxq->next_to_use += nb_rx;
if (rxq->next_to_use >= rxq->nb_rx_desc)
rxq->next_to_use = 0;
return nb_rx;
}
static inline void
hns3_rxq_rearm_mbuf_sve(struct hns3_rx_queue *rxq)
{
#define REARM_LOOP_STEP_NUM 4
struct hns3_entry *rxep = &rxq->sw_ring[rxq->rx_rearm_start];
struct hns3_desc *rxdp = rxq->rx_ring + rxq->rx_rearm_start;
struct hns3_entry *rxep_tmp = rxep;
int i;
if (unlikely(rte_mempool_get_bulk(rxq->mb_pool, (void *)rxep,
HNS3_DEFAULT_RXQ_REARM_THRESH) < 0)) {
rte_eth_devices[rxq->port_id].data->rx_mbuf_alloc_failed++;
return;
}
for (i = 0; i < HNS3_DEFAULT_RXQ_REARM_THRESH; i += REARM_LOOP_STEP_NUM,
rxep_tmp += REARM_LOOP_STEP_NUM) {
svuint64_t prf = svld1_u64(PG64_256BIT, (uint64_t *)rxep_tmp);
svprfd_gather_u64base(PG64_256BIT, prf, SV_PLDL1STRM);
}
for (i = 0; i < HNS3_DEFAULT_RXQ_REARM_THRESH; i += REARM_LOOP_STEP_NUM,
rxep += REARM_LOOP_STEP_NUM, rxdp += REARM_LOOP_STEP_NUM) {
uint64_t iova[REARM_LOOP_STEP_NUM];
iova[0] = rxep[0].mbuf->buf_iova;
iova[1] = rxep[1].mbuf->buf_iova;
iova[2] = rxep[2].mbuf->buf_iova;
iova[3] = rxep[3].mbuf->buf_iova;
svuint64_t siova = svld1_u64(PG64_256BIT, iova);
siova = svadd_n_u64_z(PG64_256BIT, siova, RTE_PKTMBUF_HEADROOM);
svuint64_t ol_base = svdup_n_u64(0);
svst1_scatter_u64offset_u64(PG64_256BIT,
(uint64_t *)&rxdp[0].addr,
svindex_u64(BD_FIELD_ADDR_OFFSET, BD_SIZE), siova);
svst1_scatter_u64offset_u64(PG64_256BIT,
(uint64_t *)&rxdp[0].addr,
svindex_u64(BD_FIELD_OL_OFFSET, BD_SIZE), ol_base);
}
rxq->rx_rearm_start += HNS3_DEFAULT_RXQ_REARM_THRESH;
if (rxq->rx_rearm_start >= rxq->nb_rx_desc)
rxq->rx_rearm_start = 0;
rxq->rx_rearm_nb -= HNS3_DEFAULT_RXQ_REARM_THRESH;
hns3_write_reg_opt(rxq->io_head_reg, HNS3_DEFAULT_RXQ_REARM_THRESH);
}
uint16_t
hns3_recv_pkts_vec_sve(void *__restrict rx_queue,
struct rte_mbuf **__restrict rx_pkts,
uint16_t nb_pkts)
{
struct hns3_rx_queue *rxq = rx_queue;
struct hns3_desc *rxdp = &rxq->rx_ring[rxq->next_to_use];
uint64_t bd_err_mask; /* bit mask indicate whick pkts is error */
uint16_t nb_rx;
rte_prefetch_non_temporal(rxdp);
nb_pkts = RTE_MIN(nb_pkts, HNS3_DEFAULT_RX_BURST);
nb_pkts = RTE_ALIGN_FLOOR(nb_pkts, HNS3_SVE_DEFAULT_DESCS_PER_LOOP);
if (rxq->rx_rearm_nb > HNS3_DEFAULT_RXQ_REARM_THRESH)
hns3_rxq_rearm_mbuf_sve(rxq);
if (unlikely(!(rxdp->rx.bd_base_info &
rte_cpu_to_le_32(1u << HNS3_RXD_VLD_B))))
return 0;
hns3_rx_prefetch_mbuf_sve(&rxq->sw_ring[rxq->next_to_use]);
bd_err_mask = 0;
nb_rx = hns3_recv_burst_vec_sve(rxq, rx_pkts, nb_pkts, &bd_err_mask);
if (unlikely(bd_err_mask))
nb_rx = hns3_rx_reassemble_pkts(rx_pkts, nb_rx, bd_err_mask);
return nb_rx;
}
static inline void
hns3_tx_free_buffers_sve(struct hns3_tx_queue *txq)
{
#define HNS3_SVE_CHECK_DESCS_PER_LOOP 8
#define TX_VLD_U8_ZIP_INDEX svindex_u8(0, 4)
svbool_t pg32 = svwhilelt_b32(0, HNS3_SVE_CHECK_DESCS_PER_LOOP);
svuint32_t vld, vld2;
svuint8_t vld_u8;
uint64_t vld_all;
struct hns3_desc *tx_desc;
int i;
/*
* All mbufs can be released only when the VLD bits of all
* descriptors in a batch are cleared.
*/
/* do logical OR operation for all desc's valid field */
vld = svdup_n_u32(0);
tx_desc = &txq->tx_ring[txq->next_to_clean];
for (i = 0; i < txq->tx_rs_thresh; i += HNS3_SVE_CHECK_DESCS_PER_LOOP,
tx_desc += HNS3_SVE_CHECK_DESCS_PER_LOOP) {
vld2 = svld1_gather_u32offset_u32(pg32, (uint32_t *)tx_desc,
svindex_u32(BD_FIELD_VALID_OFFSET, BD_SIZE));
vld = svorr_u32_z(pg32, vld, vld2);
}
/* shift left and then right to get all valid bit */
vld = svlsl_n_u32_z(pg32, vld,
HNS3_UINT32_BIT - 1 - HNS3_TXD_VLD_B);
vld = svreinterpret_u32_s32(svasr_n_s32_z(pg32,
svreinterpret_s32_u32(vld), HNS3_UINT32_BIT - 1));
/* use tbl to compress 32bit-lane to 8bit-lane */
vld_u8 = svtbl_u8(svreinterpret_u8_u32(vld), TX_VLD_U8_ZIP_INDEX);
/* dump compressed 64bit to variable */
svst1_u64(PG64_64BIT, &vld_all, svreinterpret_u64_u8(vld_u8));
if (vld_all > 0)
return;
hns3_tx_bulk_free_buffers(txq);
}
static inline void
hns3_tx_fill_hw_ring_sve(struct hns3_tx_queue *txq,
struct rte_mbuf **pkts,
uint16_t nb_pkts)
{
#define DATA_OFF_LEN_VAL_MASK 0xFFFF
struct hns3_desc *txdp = &txq->tx_ring[txq->next_to_use];
struct hns3_entry *tx_entry = &txq->sw_ring[txq->next_to_use];
const uint64_t valid_bit = (BIT(HNS3_TXD_VLD_B) | BIT(HNS3_TXD_FE_B)) <<
HNS3_UINT32_BIT;
svuint64_t base_addr, buf_iova, data_off, data_len, addr;
svuint64_t offsets = svindex_u64(0, BD_SIZE);
uint32_t i = 0;
svbool_t pg = svwhilelt_b64_u32(i, nb_pkts);
do {
base_addr = svld1_u64(pg, (uint64_t *)pkts);
/* calc mbuf's field buf_iova address */
buf_iova = svadd_n_u64_z(pg, base_addr,
offsetof(struct rte_mbuf, buf_iova));
/* calc mbuf's field data_off address */
data_off = svadd_n_u64_z(pg, base_addr,
offsetof(struct rte_mbuf, data_off));
/* calc mbuf's field data_len address */
data_len = svadd_n_u64_z(pg, base_addr,
offsetof(struct rte_mbuf, data_len));
/* store mbuf to tx_entry */
svst1_u64(pg, (uint64_t *)tx_entry, base_addr);
/* read pkts->buf_iova */
buf_iova = svld1_gather_u64base_u64(pg, buf_iova);
/* read pkts->data_off's 64bit val */
data_off = svld1_gather_u64base_u64(pg, data_off);
/* read pkts->data_len's 64bit val */
data_len = svld1_gather_u64base_u64(pg, data_len);
/* zero data_off high 48bit by svand ops */
data_off = svand_n_u64_z(pg, data_off, DATA_OFF_LEN_VAL_MASK);
/* zero data_len high 48bit by svand ops */
data_len = svand_n_u64_z(pg, data_len, DATA_OFF_LEN_VAL_MASK);
/* calc mbuf data region iova addr */
addr = svadd_u64_z(pg, buf_iova, data_off);
/* shift due data_len's offset is 2byte of BD's second 8byte */
data_len = svlsl_n_u64_z(pg, data_len, HNS3_UINT16_BIT);
/* save offset 0~7byte of every BD */
svst1_scatter_u64offset_u64(pg, (uint64_t *)&txdp->addr,
offsets, addr);
/* save offset 8~15byte of every BD */
svst1_scatter_u64offset_u64(pg, (uint64_t *)&txdp->tx.vlan_tag,
offsets, data_len);
/* save offset 16~23byte of every BD */
svst1_scatter_u64offset_u64(pg,
(uint64_t *)&txdp->tx.outer_vlan_tag,
offsets, svdup_n_u64(0));
/* save offset 24~31byte of every BD */
svst1_scatter_u64offset_u64(pg, (uint64_t *)&txdp->tx.paylen,
offsets, svdup_n_u64(valid_bit));
/* update index for next loop */
i += svcntd();
pkts += svcntd();
txdp += svcntd();
tx_entry += svcntd();
pg = svwhilelt_b64_u32(i, nb_pkts);
} while (svptest_any(svptrue_b64(), pg));
}
static uint16_t
hns3_xmit_fixed_burst_vec_sve(void *__restrict tx_queue,
struct rte_mbuf **__restrict tx_pkts,
uint16_t nb_pkts)
{
struct hns3_tx_queue *txq = (struct hns3_tx_queue *)tx_queue;
uint16_t nb_tx = 0;
if (txq->tx_bd_ready < txq->tx_free_thresh)
hns3_tx_free_buffers_sve(txq);
nb_pkts = RTE_MIN(txq->tx_bd_ready, nb_pkts);
if (unlikely(nb_pkts == 0)) {
txq->queue_full_cnt++;
return 0;
}
if (txq->next_to_use + nb_pkts > txq->nb_tx_desc) {
nb_tx = txq->nb_tx_desc - txq->next_to_use;
hns3_tx_fill_hw_ring_sve(txq, tx_pkts, nb_tx);
txq->next_to_use = 0;
}
hns3_tx_fill_hw_ring_sve(txq, tx_pkts + nb_tx, nb_pkts - nb_tx);
txq->next_to_use += nb_pkts - nb_tx;
txq->tx_bd_ready -= nb_pkts;
hns3_write_reg_opt(txq->io_tail_reg, nb_pkts);
return nb_pkts;
}
uint16_t
hns3_xmit_pkts_vec_sve(void *tx_queue,
struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
{
struct hns3_tx_queue *txq = (struct hns3_tx_queue *)tx_queue;
uint16_t ret, new_burst;
uint16_t nb_tx = 0;
while (nb_pkts) {
new_burst = RTE_MIN(nb_pkts, txq->tx_rs_thresh);
ret = hns3_xmit_fixed_burst_vec_sve(tx_queue, &tx_pkts[nb_tx],
new_burst);
nb_tx += ret;
nb_pkts -= ret;
if (ret < new_burst)
break;
}
return nb_tx;
}