#include <rte_eal.h>
#include <rte_alarm.h>
#include <rte_ether.h>
-#include <rte_ethdev.h>
-#include <rte_atomic.h>
+#include <rte_ethdev_driver.h>
#include <rte_malloc.h>
#include <rte_random.h>
#include <rte_dev.h>
struct rte_mbuf *m = mbuf;
for (; m; m = m->next, addr++) {
- *addr = m->buf_physaddr + rte_pktmbuf_headroom(m);
+ *addr = m->buf_iova + rte_pktmbuf_headroom(m);
if (*addr == 0)
goto out_err;
}
return fl->avail - fl->pend_cred <= s->fl_starve_thres;
}
+static inline unsigned int get_buf_size(struct adapter *adapter,
+ const struct rx_sw_desc *d)
+{
+ unsigned int rx_buf_size_idx = d->dma_addr & RX_BUF_SIZE;
+ unsigned int buf_size = 0;
+
+ switch (rx_buf_size_idx) {
+ case RX_SMALL_MTU_BUF:
+ buf_size = FL_MTU_SMALL_BUFSIZE(adapter);
+ break;
+
+ case RX_LARGE_MTU_BUF:
+ buf_size = FL_MTU_LARGE_BUFSIZE(adapter);
+ break;
+
+ default:
+ BUG_ON(1);
+ /* NOT REACHED */
+ }
+
+ return buf_size;
+}
+
/**
* free_rx_bufs - free the Rx buffers on an SGE free list
* @q: the SGE free list to free buffers from
* mechanism.
*/
if (unlikely(!q->bar2_addr)) {
- t4_write_reg(adap, MYPF_REG(A_SGE_PF_KDOORBELL),
- val | V_QID(q->cntxt_id));
+ t4_write_reg_relaxed(adap, MYPF_REG(A_SGE_PF_KDOORBELL),
+ val | V_QID(q->cntxt_id));
} else {
- writel(val | V_QID(q->bar2_qid),
- (void *)((uintptr_t)q->bar2_addr +
- SGE_UDB_KDOORBELL));
+ writel_relaxed(val | V_QID(q->bar2_qid),
+ (void *)((uintptr_t)q->bar2_addr +
+ SGE_UDB_KDOORBELL));
/*
* This Write memory Barrier will force the write to
unsigned int buf_size_idx = RX_SMALL_MTU_BUF;
struct rte_mbuf *buf_bulk[n];
int ret, i;
+ struct rte_pktmbuf_pool_private *mbp_priv;
+ u8 jumbo_en = rxq->rspq.eth_dev->data->dev_conf.rxmode.jumbo_frame;
+
+ /* Use jumbo mtu buffers if mbuf data room size can fit jumbo data. */
+ mbp_priv = rte_mempool_get_priv(rxq->rspq.mb_pool);
+ if (jumbo_en &&
+ ((mbp_priv->mbuf_data_room_size - RTE_PKTMBUF_HEADROOM) >= 9000))
+ buf_size_idx = RX_LARGE_MTU_BUF;
ret = rte_mempool_get_bulk(rxq->rspq.mb_pool, (void *)buf_bulk, n);
if (unlikely(ret != 0)) {
}
rte_mbuf_refcnt_set(mbuf, 1);
- mbuf->data_off = RTE_PKTMBUF_HEADROOM;
+ mbuf->data_off =
+ (uint16_t)(RTE_PTR_ALIGN((char *)mbuf->buf_addr +
+ RTE_PKTMBUF_HEADROOM,
+ adap->sge.fl_align) -
+ (char *)mbuf->buf_addr);
mbuf->next = NULL;
mbuf->nb_segs = 1;
mbuf->port = rxq->rspq.port_id;
- mapping = (dma_addr_t)(mbuf->buf_physaddr + mbuf->data_off);
+ mapping = (dma_addr_t)RTE_ALIGN(mbuf->buf_iova +
+ mbuf->data_off,
+ adap->sge.fl_align);
mapping |= buf_size_idx;
*d++ = cpu_to_be64(mapping);
set_rx_sw_desc(sd, mbuf, mapping);
* Write Header (incorporated as part of the cpl_tx_pkt_lso and
* cpl_tx_pkt structures), followed by either a TX Packet Write CPL
* message or, if we're doing a Large Send Offload, an LSO CPL message
- * with an embeded TX Packet Write CPL message.
+ * with an embedded TX Packet Write CPL message.
*/
flits = sgl_len(m->nb_segs);
if (m->tso_segsz)
#define Q_IDXDIFF(q, idx) IDXDIFF((q)->pidx, (q)->idx, (q)->size)
#define R_IDXDIFF(q, idx) IDXDIFF((q)->cidx, (q)->idx, (q)->size)
+#define PIDXDIFF(head, tail, wrap) \
+ ((tail) >= (head) ? (tail) - (head) : (wrap) - (head) + (tail))
+#define P_IDXDIFF(q, idx) PIDXDIFF((q)->cidx, idx, (q)->size)
+
/**
* ring_tx_db - ring a Tx queue's doorbell
* @adap: the adapter
}
if (likely(csum_type >= TX_CSUM_TCPIP)) {
- int hdr_len = V_TXPKT_IPHDR_LEN(m->l3_len);
+ u64 hdr_len = V_TXPKT_IPHDR_LEN(m->l3_len);
int eth_hdr_len = m->l2_len;
if (CHELSIO_CHIP_VERSION(chip) <= CHELSIO_T5)
struct adapter *adap = (struct adapter *)data;
struct sge_eth_txq *txq = &adap->sge.ethtxq[0];
int i;
+ unsigned int coal_idx;
/* monitor any pending tx */
for (i = 0; i < adap->sge.max_ethqsets; i++, txq++) {
- t4_os_lock(&txq->txq_lock);
- if (txq->q.coalesce.idx) {
- if (txq->q.coalesce.idx == txq->q.last_coal_idx &&
- txq->q.pidx == txq->q.last_pidx) {
- ship_tx_pkt_coalesce_wr(adap, txq);
- } else {
- txq->q.last_coal_idx = txq->q.coalesce.idx;
- txq->q.last_pidx = txq->q.pidx;
+ if (t4_os_trylock(&txq->txq_lock)) {
+ coal_idx = txq->q.coalesce.idx;
+ if (coal_idx) {
+ if (coal_idx == txq->q.last_coal_idx &&
+ txq->q.pidx == txq->q.last_pidx) {
+ ship_tx_pkt_coalesce_wr(adap, txq);
+ } else {
+ txq->q.last_coal_idx = coal_idx;
+ txq->q.last_pidx = txq->q.pidx;
+ }
}
+ t4_os_unlock(&txq->txq_lock);
}
- t4_os_unlock(&txq->txq_lock);
}
rte_eal_alarm_set(50, tx_timer_cb, (void *)adap);
}
/* fill the pkts WR header */
wr = (void *)&q->desc[q->pidx];
- wr->op_pkd = htonl(V_FW_WR_OP(FW_ETH_TX_PKTS_WR));
+ wr->op_pkd = htonl(V_FW_WR_OP(FW_ETH_TX_PKTS2_WR));
wr_mid = V_FW_WR_LEN16(DIV_ROUND_UP(q->coalesce.flits, 2));
ndesc = flits_to_desc(q->coalesce.flits);
struct sge_txq *q = &txq->q;
unsigned int flits, ndesc;
unsigned char type = 0;
- int credits, hw_cidx = ntohs(q->stat->cidx);
- int in_use = q->pidx - hw_cidx + flits_to_desc(q->coalesce.flits);
+ int credits;
/* use coal WR type 1 when no frags are present */
type = (mbuf->nb_segs == 1) ? 1 : 0;
- if (in_use < 0)
- in_use += q->size;
-
if (unlikely(type != q->coalesce.type && q->coalesce.idx))
ship_tx_pkt_coalesce_wr(adap, txq);
struct rte_mbuf *mbuf,
int flits, struct adapter *adap,
const struct port_info *pi,
- dma_addr_t *addr)
+ dma_addr_t *addr, uint16_t nb_pkts)
{
u64 cntrl, *end;
struct sge_txq *q = &txq->q;
struct tx_sw_desc *sd;
unsigned int idx = q->coalesce.idx, len = mbuf->pkt_len;
+#ifdef RTE_LIBRTE_CXGBE_TPUT
+ RTE_SET_USED(nb_pkts);
+#endif
+
if (q->coalesce.type == 0) {
mc = (struct ulp_txpkt *)q->coalesce.ptr;
mc->cmd_dest = htonl(V_ULPTX_CMD(4) | V_ULP_TXPKT_DEST(0) |
sd->coalesce.idx = (idx & 1) + 1;
/* send the coaelsced work request if max reached */
- if (++q->coalesce.idx == ETH_COALESCE_PKT_NUM)
+ if (++q->coalesce.idx == ETH_COALESCE_PKT_NUM
+#ifndef RTE_LIBRTE_CXGBE_TPUT
+ || q->coalesce.idx >= nb_pkts
+#endif
+ )
ship_tx_pkt_coalesce_wr(adap, txq);
return 0;
}
*
* Add a packet to an SGE Ethernet Tx queue. Runs with softirqs disabled.
*/
-int t4_eth_xmit(struct sge_eth_txq *txq, struct rte_mbuf *mbuf)
+int t4_eth_xmit(struct sge_eth_txq *txq, struct rte_mbuf *mbuf,
+ uint16_t nb_pkts)
{
const struct port_info *pi;
struct cpl_tx_pkt_lso_core *lso;
u32 wr_mid;
u64 cntrl, *end;
bool v6;
- u32 max_pkt_len = txq->eth_dev->data->dev_conf.rxmode.max_rx_pkt_len;
+ u32 max_pkt_len = txq->data->dev_conf.rxmode.max_rx_pkt_len;
/* Reject xmit if queue is stopped */
if (unlikely(txq->flags & EQ_STOPPED))
(unlikely(m->pkt_len > max_pkt_len)))
goto out_free;
- pi = (struct port_info *)txq->eth_dev->data->dev_private;
+ pi = (struct port_info *)txq->data->dev_private;
adap = pi->adapter;
cntrl = F_TXPKT_L4CSUM_DIS | F_TXPKT_IPCSUM_DIS;
}
rte_prefetch0((volatile void *)addr);
return tx_do_packet_coalesce(txq, mbuf, cflits, adap,
- pi, addr);
+ pi, addr, nb_pkts);
} else {
return -EBUSY;
}
else
lso->len = htonl(V_LSO_T5_XFER_SIZE(m->pkt_len));
cpl = (void *)(lso + 1);
- cntrl = V_TXPKT_CSUM_TYPE(v6 ? TX_CSUM_TCPIP6 : TX_CSUM_TCPIP) |
- V_TXPKT_IPHDR_LEN(l3hdr_len) |
- V_TXPKT_ETHHDR_LEN(eth_xtra_len);
+
+ if (CHELSIO_CHIP_VERSION(adap->params.chip) <= CHELSIO_T5)
+ cntrl = V_TXPKT_ETHHDR_LEN(eth_xtra_len);
+ else
+ cntrl = V_T6_TXPKT_ETHHDR_LEN(eth_xtra_len);
+
+ cntrl |= V_TXPKT_CSUM_TYPE(v6 ? TX_CSUM_TCPIP6 :
+ TX_CSUM_TCPIP) |
+ V_TXPKT_IPHDR_LEN(l3hdr_len);
txq->stats.tso++;
txq->stats.tx_cso += m->tso_segsz;
}
if (metadata)
*(void **)metadata = s;
- *phys = (uint64_t)tz->phys_addr;
+ *phys = (uint64_t)tz->iova;
return tz->addr;
}
return t4_pktgl_to_mbuf_usembufs(gl);
}
-#define RTE_MBUF_DATA_DMA_ADDR_DEFAULT(mb) \
- ((dma_addr_t) ((mb)->buf_physaddr + (mb)->data_off))
-
/**
* t4_ethrx_handler - process an ingress ethernet packet
* @q: the response queue that received the packet
const struct rss_header *rss_hdr;
bool csum_ok;
struct sge_eth_rxq *rxq = container_of(q, struct sge_eth_rxq, rspq);
+ u16 err_vec;
rss_hdr = (const void *)rsp;
pkt = (const void *)&rsp[1];
- csum_ok = pkt->csum_calc && !pkt->err_vec;
+ /* Compressed error vector is enabled for T6 only */
+ if (q->adapter->params.tp.rx_pkt_encap)
+ err_vec = G_T6_COMPR_RXERR_VEC(ntohs(pkt->err_vec));
+ else
+ err_vec = ntohs(pkt->err_vec);
+ csum_ok = pkt->csum_calc && !err_vec;
mbuf = t4_pktgl_to_mbuf(si);
if (unlikely(!mbuf)) {
}
if (pkt->vlan_ex) {
- mbuf->ol_flags |= PKT_RX_VLAN_PKT;
+ mbuf->ol_flags |= PKT_RX_VLAN;
mbuf->vlan_tci = ntohs(pkt->vlan);
}
rxq->stats.pkts++;
return 0;
}
-/**
- * is_new_response - check if a response is newly written
- * @r: the response descriptor
- * @q: the response queue
- *
- * Returns true if a response descriptor contains a yet unprocessed
- * response.
- */
-static inline bool is_new_response(const struct rsp_ctrl *r,
- const struct sge_rspq *q)
-{
- return (r->u.type_gen >> S_RSPD_GEN) == q->gen;
-}
-
#define CXGB4_MSG_AN ((void *)1)
/**
struct sge_eth_rxq *rxq = container_of(q, struct sge_eth_rxq, rspq);
while (likely(budget_left)) {
+ if (q->cidx == ntohs(q->stat->pidx))
+ break;
+
rc = (const struct rsp_ctrl *)
((const char *)q->cur_desc + (q->iqe_len - sizeof(*rc)));
- if (!is_new_response(rc, q))
- break;
-
/*
* Ensure response has been read
*/
rsp_type = G_RSPD_TYPE(rc->u.type_gen);
if (likely(rsp_type == X_RSPD_TYPE_FLBUF)) {
- const struct rx_sw_desc *rsd =
- &rxq->fl.sdesc[rxq->fl.cidx];
- const struct rss_header *rss_hdr =
- (const void *)q->cur_desc;
- const struct cpl_rx_pkt *cpl =
- (const void *)&q->cur_desc[1];
- bool csum_ok = cpl->csum_calc && !cpl->err_vec;
- struct rte_mbuf *pkt;
- u32 len = ntohl(rc->pldbuflen_qid);
-
- BUG_ON(!(len & F_RSPD_NEWBUF));
- pkt = rsd->buf;
- pkt->data_len = G_RSPD_LEN(len);
- pkt->pkt_len = pkt->data_len;
- unmap_rx_buf(&rxq->fl);
-
- if (cpl->l2info & htonl(F_RXF_IP)) {
- pkt->packet_type = RTE_PTYPE_L3_IPV4;
- if (unlikely(!csum_ok))
- pkt->ol_flags |= PKT_RX_IP_CKSUM_BAD;
-
- if ((cpl->l2info &
- htonl(F_RXF_UDP | F_RXF_TCP)) && !csum_ok)
- pkt->ol_flags |= PKT_RX_L4_CKSUM_BAD;
- } else if (cpl->l2info & htonl(F_RXF_IP6)) {
- pkt->packet_type = RTE_PTYPE_L3_IPV6;
- }
-
- if (!rss_hdr->filter_tid && rss_hdr->hash_type) {
- pkt->ol_flags |= PKT_RX_RSS_HASH;
- pkt->hash.rss = ntohl(rss_hdr->hash_val);
- }
-
- if (cpl->vlan_ex) {
- pkt->ol_flags |= PKT_RX_VLAN_PKT;
- pkt->vlan_tci = ntohs(cpl->vlan);
+ unsigned int stat_pidx;
+ int stat_pidx_diff;
+
+ stat_pidx = ntohs(q->stat->pidx);
+ stat_pidx_diff = P_IDXDIFF(q, stat_pidx);
+ while (stat_pidx_diff && budget_left) {
+ const struct rx_sw_desc *rsd =
+ &rxq->fl.sdesc[rxq->fl.cidx];
+ const struct rss_header *rss_hdr =
+ (const void *)q->cur_desc;
+ const struct cpl_rx_pkt *cpl =
+ (const void *)&q->cur_desc[1];
+ struct rte_mbuf *pkt, *npkt;
+ u32 len, bufsz;
+ bool csum_ok;
+ u16 err_vec;
+
+ rc = (const struct rsp_ctrl *)
+ ((const char *)q->cur_desc +
+ (q->iqe_len - sizeof(*rc)));
+
+ rsp_type = G_RSPD_TYPE(rc->u.type_gen);
+ if (unlikely(rsp_type != X_RSPD_TYPE_FLBUF))
+ break;
+
+ len = ntohl(rc->pldbuflen_qid);
+ BUG_ON(!(len & F_RSPD_NEWBUF));
+ pkt = rsd->buf;
+ npkt = pkt;
+ len = G_RSPD_LEN(len);
+ pkt->pkt_len = len;
+
+ /* Compressed error vector is enabled for
+ * T6 only
+ */
+ if (q->adapter->params.tp.rx_pkt_encap)
+ err_vec = G_T6_COMPR_RXERR_VEC(
+ ntohs(cpl->err_vec));
+ else
+ err_vec = ntohs(cpl->err_vec);
+ csum_ok = cpl->csum_calc && !err_vec;
+
+ /* Chain mbufs into len if necessary */
+ while (len) {
+ struct rte_mbuf *new_pkt = rsd->buf;
+
+ bufsz = min(get_buf_size(q->adapter,
+ rsd), len);
+ new_pkt->data_len = bufsz;
+ unmap_rx_buf(&rxq->fl);
+ len -= bufsz;
+ npkt->next = new_pkt;
+ npkt = new_pkt;
+ pkt->nb_segs++;
+ rsd = &rxq->fl.sdesc[rxq->fl.cidx];
+ }
+ npkt->next = NULL;
+ pkt->nb_segs--;
+
+ if (cpl->l2info & htonl(F_RXF_IP)) {
+ pkt->packet_type = RTE_PTYPE_L3_IPV4;
+ if (unlikely(!csum_ok))
+ pkt->ol_flags |=
+ PKT_RX_IP_CKSUM_BAD;
+
+ if ((cpl->l2info &
+ htonl(F_RXF_UDP | F_RXF_TCP)) &&
+ !csum_ok)
+ pkt->ol_flags |=
+ PKT_RX_L4_CKSUM_BAD;
+ } else if (cpl->l2info & htonl(F_RXF_IP6)) {
+ pkt->packet_type = RTE_PTYPE_L3_IPV6;
+ }
+
+ if (!rss_hdr->filter_tid &&
+ rss_hdr->hash_type) {
+ pkt->ol_flags |= PKT_RX_RSS_HASH;
+ pkt->hash.rss =
+ ntohl(rss_hdr->hash_val);
+ }
+
+ if (cpl->vlan_ex) {
+ pkt->ol_flags |= PKT_RX_VLAN;
+ pkt->vlan_tci = ntohs(cpl->vlan);
+ }
+
+ rxq->stats.pkts++;
+ rxq->stats.rx_bytes += pkt->pkt_len;
+ rx_pkts[budget - budget_left] = pkt;
+
+ rspq_next(q);
+ budget_left--;
+ stat_pidx_diff--;
}
- rxq->stats.pkts++;
- rxq->stats.rx_bytes += pkt->pkt_len;
- rx_pkts[budget - budget_left] = pkt;
+ continue;
} else if (likely(rsp_type == X_RSPD_TYPE_CPL)) {
ret = q->handler(q, q->cur_desc, NULL);
} else {
rspq_next(q);
budget_left--;
-
- if (R_IDXDIFF(q, gts_idx) >= 64) {
- unsigned int cidx_inc = R_IDXDIFF(q, gts_idx);
- unsigned int params;
- u32 val;
-
- if (fl_cap(&rxq->fl) - rxq->fl.avail >= 64)
- __refill_fl(q->adapter, &rxq->fl);
- params = V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX);
- q->next_intr_params = params;
- val = V_CIDXINC(cidx_inc) | V_SEINTARM(params);
-
- if (unlikely(!q->bar2_addr))
- t4_write_reg(q->adapter, MYPF_REG(A_SGE_PF_GTS),
- val |
- V_INGRESSQID((u32)q->cntxt_id));
- else {
- writel(val | V_INGRESSQID(q->bar2_qid),
- (void *)((uintptr_t)q->bar2_addr +
- SGE_UDB_GTS));
- /*
- * This Write memory Barrier will force the
- * write to the User Doorbell area to be
- * flushed.
- */
- wmb();
- }
- q->gts_idx = q->cidx;
- }
}
/*
int cxgbe_poll(struct sge_rspq *q, struct rte_mbuf **rx_pkts,
unsigned int budget, unsigned int *work_done)
{
- int err = 0;
+ struct sge_eth_rxq *rxq = container_of(q, struct sge_eth_rxq, rspq);
+ unsigned int cidx_inc;
+ unsigned int params;
+ u32 val;
*work_done = process_responses(q, budget, rx_pkts);
- return err;
+
+ if (*work_done) {
+ cidx_inc = R_IDXDIFF(q, gts_idx);
+
+ if (q->offset >= 0 && fl_cap(&rxq->fl) - rxq->fl.avail >= 64)
+ __refill_fl(q->adapter, &rxq->fl);
+
+ params = q->intr_params;
+ q->next_intr_params = params;
+ val = V_CIDXINC(cidx_inc) | V_SEINTARM(params);
+
+ if (unlikely(!q->bar2_addr)) {
+ t4_write_reg(q->adapter, MYPF_REG(A_SGE_PF_GTS),
+ val | V_INGRESSQID((u32)q->cntxt_id));
+ } else {
+ writel(val | V_INGRESSQID(q->bar2_qid),
+ (void *)((uintptr_t)q->bar2_addr + SGE_UDB_GTS));
+ /* This Write memory Barrier will force the
+ * write to the User Doorbell area to be
+ * flushed.
+ */
+ wmb();
+ }
+ q->gts_idx = q->cidx;
+ }
+ return 0;
}
/**
char z_name[RTE_MEMZONE_NAMESIZE];
char z_name_sw[RTE_MEMZONE_NAMESIZE];
unsigned int nb_refill;
+ u8 pciechan;
/* Size needs to be multiple of 16, including status entry. */
iq->size = cxgbe_roundup(iq->size, 16);
snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
- eth_dev->driver->pci_drv.name, fwevtq ? "fwq_ring" : "rx_ring",
+ eth_dev->device->driver->name,
+ fwevtq ? "fwq_ring" : "rx_ring",
eth_dev->data->port_id, queue_id);
snprintf(z_name_sw, sizeof(z_name_sw), "%s_sw_ring", z_name);
memset(&c, 0, sizeof(c));
c.op_to_vfn = htonl(V_FW_CMD_OP(FW_IQ_CMD) | F_FW_CMD_REQUEST |
- F_FW_CMD_WRITE | F_FW_CMD_EXEC |
- V_FW_IQ_CMD_PFN(adap->pf) | V_FW_IQ_CMD_VFN(0));
+ F_FW_CMD_WRITE | F_FW_CMD_EXEC);
+
+ if (is_pf4(adap)) {
+ pciechan = cong > 0 ? cxgbe_ffs(cong) - 1 : pi->tx_chan;
+ c.op_to_vfn |= htonl(V_FW_IQ_CMD_PFN(adap->pf) |
+ V_FW_IQ_CMD_VFN(0));
+ if (cong >= 0)
+ c.iqns_to_fl0congen = htonl(F_FW_IQ_CMD_IQFLINTCONGEN |
+ F_FW_IQ_CMD_IQRO);
+ } else {
+ pciechan = pi->port_id;
+ }
+
c.alloc_to_len16 = htonl(F_FW_IQ_CMD_ALLOC | F_FW_IQ_CMD_IQSTART |
(sizeof(c) / 16));
c.type_to_iqandstindex =
V_FW_IQ_CMD_IQASYNCH(fwevtq) |
V_FW_IQ_CMD_VIID(pi->viid) |
V_FW_IQ_CMD_IQANDST(intr_idx < 0) |
- V_FW_IQ_CMD_IQANUD(X_UPDATEDELIVERY_INTERRUPT) |
+ V_FW_IQ_CMD_IQANUD(X_UPDATEDELIVERY_STATUS_PAGE) |
V_FW_IQ_CMD_IQANDSTINDEX(intr_idx >= 0 ? intr_idx :
-intr_idx - 1));
c.iqdroprss_to_iqesize =
- htons(V_FW_IQ_CMD_IQPCIECH(pi->tx_chan) |
+ htons(V_FW_IQ_CMD_IQPCIECH(pciechan) |
F_FW_IQ_CMD_IQGTSMODE |
V_FW_IQ_CMD_IQINTCNTTHRESH(iq->pktcnt_idx) |
V_FW_IQ_CMD_IQESIZE(ilog2(iq->iqe_len) - 4));
c.iqsize = htons(iq->size);
c.iqaddr = cpu_to_be64(iq->phys_addr);
- if (cong >= 0)
- c.iqns_to_fl0congen = htonl(F_FW_IQ_CMD_IQFLINTCONGEN);
if (fl) {
struct sge_eth_rxq *rxq = container_of(fl, struct sge_eth_rxq,
fl);
- enum chip_type chip = (enum chip_type)CHELSIO_CHIP_VERSION(
- adap->params.chip);
+ unsigned int chip_ver = CHELSIO_CHIP_VERSION(adap->params.chip);
/*
* Allocate the ring for the hardware free list (with space
fl->size = cxgbe_roundup(fl->size, 8);
snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
- eth_dev->driver->pci_drv.name,
+ eth_dev->device->driver->name,
fwevtq ? "fwq_ring" : "fl_ring",
eth_dev->data->port_id, queue_id);
snprintf(z_name_sw, sizeof(z_name_sw), "%s_sw_ring", z_name);
0 : F_FW_IQ_CMD_FL0PACKEN) |
F_FW_IQ_CMD_FL0FETCHRO | F_FW_IQ_CMD_FL0DATARO |
F_FW_IQ_CMD_FL0PADEN);
- if (cong >= 0)
+ if (is_pf4(adap) && cong >= 0)
c.iqns_to_fl0congen |=
htonl(V_FW_IQ_CMD_FL0CNGCHMAP(cong) |
F_FW_IQ_CMD_FL0CONGCIF |
* Hence maximum allowed burst size will be 448 bytes.
*/
c.fl0dcaen_to_fl0cidxfthresh =
- htons(V_FW_IQ_CMD_FL0FBMIN(X_FETCHBURSTMIN_128B) |
- V_FW_IQ_CMD_FL0FBMAX((chip <= CHELSIO_T5) ?
- X_FETCHBURSTMAX_512B : X_FETCHBURSTMAX_256B));
+ htons(V_FW_IQ_CMD_FL0FBMIN(chip_ver <= CHELSIO_T5 ?
+ X_FETCHBURSTMIN_128B :
+ X_FETCHBURSTMIN_64B) |
+ V_FW_IQ_CMD_FL0FBMAX(chip_ver <= CHELSIO_T5 ?
+ X_FETCHBURSTMAX_512B :
+ X_FETCHBURSTMAX_256B));
c.fl0size = htons(flsz);
c.fl0addr = cpu_to_be64(fl->addr);
}
- ret = t4_wr_mbox(adap, adap->mbox, &c, sizeof(c), &c);
+ if (is_pf4(adap))
+ ret = t4_wr_mbox(adap, adap->mbox, &c, sizeof(c), &c);
+ else
+ ret = t4vf_wr_mbox(adap, &c, sizeof(c), &c);
if (ret)
goto err;
iq->bar2_addr = bar2_address(adap, iq->cntxt_id, T4_BAR2_QTYPE_INGRESS,
&iq->bar2_qid);
iq->size--; /* subtract status entry */
+ iq->stat = (void *)&iq->desc[iq->size * 8];
iq->eth_dev = eth_dev;
iq->handler = hnd;
- iq->port_id = pi->port_id;
+ iq->port_id = pi->pidx;
iq->mb_pool = mp;
/* set offset to -1 to distinguish ingress queues without FL */
* a lot easier to fix in one place ... For now we do something very
* simple (and hopefully less wrong).
*/
- if (!is_t4(adap->params.chip) && cong >= 0) {
+ if (is_pf4(adap) && !is_t4(adap->params.chip) && cong >= 0) {
u32 param, val;
int i;
refill_fl_err:
t4_iq_free(adap, adap->mbox, adap->pf, 0, FW_IQ_TYPE_FL_INT_CAP,
- iq->cntxt_id, fl ? fl->cntxt_id : 0xffff, 0xffff);
+ iq->cntxt_id, fl->cntxt_id, 0xffff);
fl_nomem:
ret = -ENOMEM;
err:
return ret;
}
-static void init_txq(struct adapter *adap, struct sge_txq *q, unsigned int id)
+static void init_txq(struct adapter *adap, struct sge_txq *q, unsigned int id,
+ unsigned int abs_id)
{
q->cntxt_id = id;
+ q->abs_id = abs_id;
q->bar2_addr = bar2_address(adap, q->cntxt_id, T4_BAR2_QTYPE_EGRESS,
&q->bar2_qid);
q->cidx = 0;
struct port_info *pi = (struct port_info *)(eth_dev->data->dev_private);
char z_name[RTE_MEMZONE_NAMESIZE];
char z_name_sw[RTE_MEMZONE_NAMESIZE];
+ u8 pciechan;
/* Add status entries */
nentries = txq->q.size + s->stat_len / sizeof(struct tx_desc);
snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
- eth_dev->driver->pci_drv.name, "tx_ring",
+ eth_dev->device->driver->name, "tx_ring",
eth_dev->data->port_id, queue_id);
snprintf(z_name_sw, sizeof(z_name_sw), "%s_sw_ring", z_name);
memset(&c, 0, sizeof(c));
c.op_to_vfn = htonl(V_FW_CMD_OP(FW_EQ_ETH_CMD) | F_FW_CMD_REQUEST |
- F_FW_CMD_WRITE | F_FW_CMD_EXEC |
- V_FW_EQ_ETH_CMD_PFN(adap->pf) |
- V_FW_EQ_ETH_CMD_VFN(0));
+ F_FW_CMD_WRITE | F_FW_CMD_EXEC);
+ if (is_pf4(adap)) {
+ pciechan = pi->tx_chan;
+ c.op_to_vfn |= htonl(V_FW_EQ_ETH_CMD_PFN(adap->pf) |
+ V_FW_EQ_ETH_CMD_VFN(0));
+ } else {
+ pciechan = pi->port_id;
+ }
+
c.alloc_to_len16 = htonl(F_FW_EQ_ETH_CMD_ALLOC |
F_FW_EQ_ETH_CMD_EQSTART | (sizeof(c) / 16));
c.autoequiqe_to_viid = htonl(F_FW_EQ_ETH_CMD_AUTOEQUEQE |
V_FW_EQ_ETH_CMD_VIID(pi->viid));
c.fetchszm_to_iqid =
htonl(V_FW_EQ_ETH_CMD_HOSTFCMODE(X_HOSTFCMODE_NONE) |
- V_FW_EQ_ETH_CMD_PCIECHN(pi->tx_chan) |
+ V_FW_EQ_ETH_CMD_PCIECHN(pciechan) |
F_FW_EQ_ETH_CMD_FETCHRO | V_FW_EQ_ETH_CMD_IQID(iqid));
c.dcaen_to_eqsize =
htonl(V_FW_EQ_ETH_CMD_FBMIN(X_FETCHBURSTMIN_64B) |
V_FW_EQ_ETH_CMD_EQSIZE(nentries));
c.eqaddr = rte_cpu_to_be_64(txq->q.phys_addr);
- ret = t4_wr_mbox(adap, adap->mbox, &c, sizeof(c), &c);
+ if (is_pf4(adap))
+ ret = t4_wr_mbox(adap, adap->mbox, &c, sizeof(c), &c);
+ else
+ ret = t4vf_wr_mbox(adap, &c, sizeof(c), &c);
if (ret) {
rte_free(txq->q.sdesc);
txq->q.sdesc = NULL;
return ret;
}
- init_txq(adap, &txq->q, G_FW_EQ_ETH_CMD_EQID(ntohl(c.eqid_pkd)));
+ init_txq(adap, &txq->q, G_FW_EQ_ETH_CMD_EQID(ntohl(c.eqid_pkd)),
+ G_FW_EQ_ETH_CMD_PHYSEQID(ntohl(c.physeqid_pkd)));
txq->stats.tso = 0;
txq->stats.pkts = 0;
txq->stats.tx_cso = 0;
txq->stats.mapping_err = 0;
txq->flags |= EQ_STOPPED;
txq->eth_dev = eth_dev;
+ txq->data = eth_dev->data;
t4_os_lock_init(&txq->txq_lock);
return 0;
}
int t4_sge_init(struct adapter *adap)
{
struct sge *s = &adap->sge;
- u32 sge_control, sge_control2, sge_conm_ctrl;
- unsigned int ingpadboundary, ingpackboundary;
+ u32 sge_control, sge_conm_ctrl;
int ret, egress_threshold;
/*
sge_control = t4_read_reg(adap, A_SGE_CONTROL);
s->pktshift = G_PKTSHIFT(sge_control);
s->stat_len = (sge_control & F_EGRSTATUSPAGESIZE) ? 128 : 64;
-
- /*
- * T4 uses a single control field to specify both the PCIe Padding and
- * Packing Boundary. T5 introduced the ability to specify these
- * separately. The actual Ingress Packet Data alignment boundary
- * within Packed Buffer Mode is the maximum of these two
- * specifications.
- */
- ingpadboundary = 1 << (G_INGPADBOUNDARY(sge_control) +
- X_INGPADBOUNDARY_SHIFT);
- s->fl_align = ingpadboundary;
-
- if (!is_t4(adap->params.chip) && !adap->use_unpacked_mode) {
- /*
- * T5 has a weird interpretation of one of the PCIe Packing
- * Boundary values. No idea why ...
- */
- sge_control2 = t4_read_reg(adap, A_SGE_CONTROL2);
- ingpackboundary = G_INGPACKBOUNDARY(sge_control2);
- if (ingpackboundary == X_INGPACKBOUNDARY_16B)
- ingpackboundary = 16;
- else
- ingpackboundary = 1 << (ingpackboundary +
- X_INGPACKBOUNDARY_SHIFT);
-
- s->fl_align = max(ingpadboundary, ingpackboundary);
- }
-
+ s->fl_align = t4_fl_pkt_align(adap);
ret = t4_sge_init_soft(adap);
if (ret < 0) {
dev_err(adap, "%s: t4_sge_init_soft failed, error %d\n",
return 0;
}
+
+int t4vf_sge_init(struct adapter *adap)
+{
+ struct sge_params *sge_params = &adap->params.sge;
+ u32 sge_ingress_queues_per_page;
+ u32 sge_egress_queues_per_page;
+ u32 sge_control, sge_control2;
+ u32 fl_small_pg, fl_large_pg;
+ u32 sge_ingress_rx_threshold;
+ u32 sge_timer_value_0_and_1;
+ u32 sge_timer_value_2_and_3;
+ u32 sge_timer_value_4_and_5;
+ u32 sge_congestion_control;
+ struct sge *s = &adap->sge;
+ unsigned int s_hps, s_qpp;
+ u32 sge_host_page_size;
+ u32 params[7], vals[7];
+ int v;
+
+ /* query basic params from fw */
+ params[0] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_CONTROL));
+ params[1] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_HOST_PAGE_SIZE));
+ params[2] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_FL_BUFFER_SIZE0));
+ params[3] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_FL_BUFFER_SIZE1));
+ params[4] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_TIMER_VALUE_0_AND_1));
+ params[5] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_TIMER_VALUE_2_AND_3));
+ params[6] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_TIMER_VALUE_4_AND_5));
+ v = t4vf_query_params(adap, 7, params, vals);
+ if (v != FW_SUCCESS)
+ return v;
+
+ sge_control = vals[0];
+ sge_host_page_size = vals[1];
+ fl_small_pg = vals[2];
+ fl_large_pg = vals[3];
+ sge_timer_value_0_and_1 = vals[4];
+ sge_timer_value_2_and_3 = vals[5];
+ sge_timer_value_4_and_5 = vals[6];
+
+ /*
+ * Start by vetting the basic SGE parameters which have been set up by
+ * the Physical Function Driver.
+ */
+
+ /* We only bother using the Large Page logic if the Large Page Buffer
+ * is larger than our Page Size Buffer.
+ */
+ if (fl_large_pg <= fl_small_pg)
+ fl_large_pg = 0;
+
+ /* The Page Size Buffer must be exactly equal to our Page Size and the
+ * Large Page Size Buffer should be 0 (per above) or a power of 2.
+ */
+ if (fl_small_pg != CXGBE_PAGE_SIZE ||
+ (fl_large_pg & (fl_large_pg - 1)) != 0) {
+ dev_err(adapter->pdev_dev, "bad SGE FL buffer sizes [%d, %d]\n",
+ fl_small_pg, fl_large_pg);
+ return -EINVAL;
+ }
+
+ if ((sge_control & F_RXPKTCPLMODE) !=
+ V_RXPKTCPLMODE(X_RXPKTCPLMODE_SPLIT)) {
+ dev_err(adapter->pdev_dev, "bad SGE CPL MODE\n");
+ return -EINVAL;
+ }
+
+
+ /* Grab ingress packing boundary from SGE_CONTROL2 for */
+ params[0] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_CONTROL2));
+ v = t4vf_query_params(adap, 1, params, vals);
+ if (v != FW_SUCCESS) {
+ dev_err(adapter, "Unable to get SGE Control2; "
+ "probably old firmware.\n");
+ return v;
+ }
+ sge_control2 = vals[0];
+
+ params[0] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_INGRESS_RX_THRESHOLD));
+ params[1] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_CONM_CTRL));
+ v = t4vf_query_params(adap, 2, params, vals);
+ if (v != FW_SUCCESS)
+ return v;
+ sge_ingress_rx_threshold = vals[0];
+ sge_congestion_control = vals[1];
+ params[0] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_EGRESS_QUEUES_PER_PAGE_VF));
+ params[1] = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_REG) |
+ V_FW_PARAMS_PARAM_XYZ(A_SGE_INGRESS_QUEUES_PER_PAGE_VF));
+ v = t4vf_query_params(adap, 2, params, vals);
+ if (v != FW_SUCCESS) {
+ dev_warn(adap, "Unable to get VF SGE Queues/Page; "
+ "probably old firmware.\n");
+ return v;
+ }
+ sge_egress_queues_per_page = vals[0];
+ sge_ingress_queues_per_page = vals[1];
+
+ /*
+ * We need the Queues/Page for our VF. This is based on the
+ * PF from which we're instantiated and is indexed in the
+ * register we just read.
+ */
+ s_hps = (S_HOSTPAGESIZEPF0 +
+ (S_HOSTPAGESIZEPF1 - S_HOSTPAGESIZEPF0) * adap->pf);
+ sge_params->hps =
+ ((sge_host_page_size >> s_hps) & M_HOSTPAGESIZEPF0);
+
+ s_qpp = (S_QUEUESPERPAGEPF0 +
+ (S_QUEUESPERPAGEPF1 - S_QUEUESPERPAGEPF0) * adap->pf);
+ sge_params->eq_qpp =
+ ((sge_egress_queues_per_page >> s_qpp)
+ & M_QUEUESPERPAGEPF0);
+ sge_params->iq_qpp =
+ ((sge_ingress_queues_per_page >> s_qpp)
+ & M_QUEUESPERPAGEPF0);
+
+ /*
+ * Now translate the queried parameters into our internal forms.
+ */
+ if (fl_large_pg)
+ s->fl_pg_order = ilog2(fl_large_pg) - PAGE_SHIFT;
+ s->stat_len = ((sge_control & F_EGRSTATUSPAGESIZE)
+ ? 128 : 64);
+ s->pktshift = G_PKTSHIFT(sge_control);
+ s->fl_align = t4vf_fl_pkt_align(adap, sge_control, sge_control2);
+
+ /*
+ * A FL with <= fl_starve_thres buffers is starving and a periodic
+ * timer will attempt to refill it. This needs to be larger than the
+ * SGE's Egress Congestion Threshold. If it isn't, then we can get
+ * stuck waiting for new packets while the SGE is waiting for us to
+ * give it more Free List entries. (Note that the SGE's Egress
+ * Congestion Threshold is in units of 2 Free List pointers.)
+ */
+ switch (CHELSIO_CHIP_VERSION(adap->params.chip)) {
+ case CHELSIO_T5:
+ s->fl_starve_thres =
+ G_EGRTHRESHOLDPACKING(sge_congestion_control);
+ break;
+ case CHELSIO_T6:
+ default:
+ s->fl_starve_thres =
+ G_T6_EGRTHRESHOLDPACKING(sge_congestion_control);
+ break;
+ }
+ s->fl_starve_thres = s->fl_starve_thres * 2 + 1;
+
+ /*
+ * Save RX interrupt holdoff timer values and counter
+ * threshold values from the SGE parameters.
+ */
+ s->timer_val[0] = core_ticks_to_us(adap,
+ G_TIMERVALUE0(sge_timer_value_0_and_1));
+ s->timer_val[1] = core_ticks_to_us(adap,
+ G_TIMERVALUE1(sge_timer_value_0_and_1));
+ s->timer_val[2] = core_ticks_to_us(adap,
+ G_TIMERVALUE2(sge_timer_value_2_and_3));
+ s->timer_val[3] = core_ticks_to_us(adap,
+ G_TIMERVALUE3(sge_timer_value_2_and_3));
+ s->timer_val[4] = core_ticks_to_us(adap,
+ G_TIMERVALUE4(sge_timer_value_4_and_5));
+ s->timer_val[5] = core_ticks_to_us(adap,
+ G_TIMERVALUE5(sge_timer_value_4_and_5));
+ s->counter_val[0] = G_THRESHOLD_0(sge_ingress_rx_threshold);
+ s->counter_val[1] = G_THRESHOLD_1(sge_ingress_rx_threshold);
+ s->counter_val[2] = G_THRESHOLD_2(sge_ingress_rx_threshold);
+ s->counter_val[3] = G_THRESHOLD_3(sge_ingress_rx_threshold);
+ return 0;
+}
git.droids-corp.org / dpdk.git / blobdiff