diff options
Diffstat (limited to 'drivers/net/ethernet/marvell/octeontx2/nic')
18 files changed, 735 insertions, 174 deletions
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/Makefile b/drivers/net/ethernet/marvell/octeontx2/nic/Makefile index cb6513ab35e7..69e0778f9ac1 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/Makefile +++ b/drivers/net/ethernet/marvell/octeontx2/nic/Makefile @@ -9,7 +9,7 @@ obj-$(CONFIG_RVU_ESWITCH) += rvu_rep.o rvu_nicpf-y := otx2_pf.o otx2_common.o otx2_txrx.o otx2_ethtool.o \ otx2_flows.o otx2_tc.o cn10k.o otx2_dmac_flt.o \ - otx2_devlink.o qos_sq.o qos.o + otx2_devlink.o qos_sq.o qos.o otx2_xsk.o rvu_nicvf-y := otx2_vf.o rvu_rep-y := rep.o diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k.c index a15cc86635d6..7f6a435ac680 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k.c @@ -112,9 +112,12 @@ int cn10k_refill_pool_ptrs(void *dev, struct otx2_cq_queue *cq) struct otx2_nic *pfvf = dev; int cnt = cq->pool_ptrs; u64 ptrs[NPA_MAX_BURST]; + struct otx2_pool *pool; dma_addr_t bufptr; int num_ptrs = 1; + pool = &pfvf->qset.pool[cq->cq_idx]; + /* Refill pool with new buffers */ while (cq->pool_ptrs) { if (otx2_alloc_buffer(pfvf, cq, &bufptr)) { @@ -124,7 +127,9 @@ int cn10k_refill_pool_ptrs(void *dev, struct otx2_cq_queue *cq) break; } cq->pool_ptrs--; - ptrs[num_ptrs] = (u64)bufptr + OTX2_HEAD_ROOM; + ptrs[num_ptrs] = pool->xsk_pool ? + (u64)bufptr : (u64)bufptr + OTX2_HEAD_ROOM; + num_ptrs++; if (num_ptrs == NPA_MAX_BURST || cq->pool_ptrs == 0) { __cn10k_aura_freeptr(pfvf, cq->cq_idx, ptrs, @@ -352,9 +357,12 @@ int cn10k_free_matchall_ipolicer(struct otx2_nic *pfvf) mutex_lock(&pfvf->mbox.lock); /* Remove RQ's policer mapping */ - for (qidx = 0; qidx < hw->rx_queues; qidx++) - cn10k_map_unmap_rq_policer(pfvf, qidx, - hw->matchall_ipolicer, false); + for (qidx = 0; qidx < hw->rx_queues; qidx++) { + rc = cn10k_map_unmap_rq_policer(pfvf, qidx, hw->matchall_ipolicer, false); + if (rc) + dev_warn(pfvf->dev, "Failed to unmap RQ %d's policer (error %d).", + qidx, rc); + } rc = cn10k_free_leaf_profile(pfvf, hw->matchall_ipolicer); diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c index 09a5b5268205..a6500e3673f2 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c @@ -663,10 +663,10 @@ static int cn10k_ipsec_inb_add_state(struct xfrm_state *x, return -EOPNOTSUPP; } -static int cn10k_ipsec_outb_add_state(struct xfrm_state *x, +static int cn10k_ipsec_outb_add_state(struct net_device *dev, + struct xfrm_state *x, struct netlink_ext_ack *extack) { - struct net_device *netdev = x->xso.dev; struct cn10k_tx_sa_s *sa_entry; struct qmem *sa_info; struct otx2_nic *pf; @@ -676,7 +676,7 @@ static int cn10k_ipsec_outb_add_state(struct xfrm_state *x, if (err) return err; - pf = netdev_priv(netdev); + pf = netdev_priv(dev); err = qmem_alloc(pf->dev, &sa_info, pf->ipsec.sa_size, OTX2_ALIGN); if (err) @@ -700,18 +700,18 @@ static int cn10k_ipsec_outb_add_state(struct xfrm_state *x, return 0; } -static int cn10k_ipsec_add_state(struct xfrm_state *x, +static int cn10k_ipsec_add_state(struct net_device *dev, + struct xfrm_state *x, struct netlink_ext_ack *extack) { if (x->xso.dir == XFRM_DEV_OFFLOAD_IN) return cn10k_ipsec_inb_add_state(x, extack); else - return cn10k_ipsec_outb_add_state(x, extack); + return cn10k_ipsec_outb_add_state(dev, x, extack); } -static void cn10k_ipsec_del_state(struct xfrm_state *x) +static void cn10k_ipsec_del_state(struct net_device *dev, struct xfrm_state *x) { - struct net_device *netdev = x->xso.dev; struct cn10k_tx_sa_s *sa_entry; struct qmem *sa_info; struct otx2_nic *pf; @@ -720,7 +720,7 @@ static void cn10k_ipsec_del_state(struct xfrm_state *x) if (x->xso.dir == XFRM_DEV_OFFLOAD_IN) return; - pf = netdev_priv(netdev); + pf = netdev_priv(dev); sa_info = (struct qmem *)x->xso.offload_handle; sa_entry = (struct cn10k_tx_sa_s *)sa_info->base; @@ -732,7 +732,7 @@ static void cn10k_ipsec_del_state(struct xfrm_state *x) err = cn10k_outb_write_sa(pf, sa_info); if (err) - netdev_err(netdev, "Error (%d) deleting SA\n", err); + netdev_err(dev, "Error (%d) deleting SA\n", err); x->xso.offload_handle = 0; qmem_free(pf->dev, sa_info); @@ -744,24 +744,9 @@ static void cn10k_ipsec_del_state(struct xfrm_state *x) queue_work(pf->ipsec.sa_workq, &pf->ipsec.sa_work); } -static bool cn10k_ipsec_offload_ok(struct sk_buff *skb, struct xfrm_state *x) -{ - if (x->props.family == AF_INET) { - /* Offload with IPv4 options is not supported yet */ - if (ip_hdr(skb)->ihl > 5) - return false; - } else { - /* Offload with IPv6 extension headers is not support yet */ - if (ipv6_ext_hdr(ipv6_hdr(skb)->nexthdr)) - return false; - } - return true; -} - static const struct xfrmdev_ops cn10k_ipsec_xfrmdev_ops = { .xdo_dev_state_add = cn10k_ipsec_add_state, .xdo_dev_state_delete = cn10k_ipsec_del_state, - .xdo_dev_offload_ok = cn10k_ipsec_offload_ok, }; static void cn10k_ipsec_sa_wq_handler(struct work_struct *work) diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c index f3b9daffaec3..4c7e0f345cb5 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c @@ -531,7 +531,8 @@ static int cn10k_mcs_write_tx_secy(struct otx2_nic *pfvf, if (sw_tx_sc->encrypt) sectag_tci |= (MCS_TCI_E | MCS_TCI_C); - policy = FIELD_PREP(MCS_TX_SECY_PLCY_MTU, secy->netdev->mtu); + policy = FIELD_PREP(MCS_TX_SECY_PLCY_MTU, + pfvf->netdev->mtu + OTX2_ETH_HLEN); /* Write SecTag excluding AN bits(1..0) */ policy |= FIELD_PREP(MCS_TX_SECY_PLCY_ST_TCI, sectag_tci >> 2); policy |= FIELD_PREP(MCS_TX_SECY_PLCY_ST_OFFSET, tag_offset); diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c index 2b49bfec7869..6b5c9536d26d 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c @@ -17,6 +17,7 @@ #include "otx2_common.h" #include "otx2_struct.h" #include "cn10k.h" +#include "otx2_xsk.h" static bool otx2_is_pfc_enabled(struct otx2_nic *pfvf) { @@ -330,6 +331,10 @@ int otx2_set_rss_table(struct otx2_nic *pfvf, int ctx_id) rss_ctx = rss->rss_ctx[ctx_id]; /* Get memory to put this msg */ for (idx = 0; idx < rss->rss_size; idx++) { + /* Ignore the queue if AF_XDP zero copy is enabled */ + if (test_bit(rss_ctx->ind_tbl[idx], pfvf->af_xdp_zc_qidx)) + continue; + aq = otx2_mbox_alloc_msg_nix_aq_enq(mbox); if (!aq) { /* The shared memory buffer can be full. @@ -549,10 +554,13 @@ static int otx2_alloc_pool_buf(struct otx2_nic *pfvf, struct otx2_pool *pool, } static int __otx2_alloc_rbuf(struct otx2_nic *pfvf, struct otx2_pool *pool, - dma_addr_t *dma) + dma_addr_t *dma, int qidx, int idx) { u8 *buf; + if (pool->xsk_pool) + return otx2_xsk_pool_alloc_buf(pfvf, pool, dma, idx); + if (pool->page_pool) return otx2_alloc_pool_buf(pfvf, pool, dma); @@ -571,12 +579,12 @@ static int __otx2_alloc_rbuf(struct otx2_nic *pfvf, struct otx2_pool *pool, } int otx2_alloc_rbuf(struct otx2_nic *pfvf, struct otx2_pool *pool, - dma_addr_t *dma) + dma_addr_t *dma, int qidx, int idx) { int ret; local_bh_disable(); - ret = __otx2_alloc_rbuf(pfvf, pool, dma); + ret = __otx2_alloc_rbuf(pfvf, pool, dma, qidx, idx); local_bh_enable(); return ret; } @@ -584,7 +592,8 @@ int otx2_alloc_rbuf(struct otx2_nic *pfvf, struct otx2_pool *pool, int otx2_alloc_buffer(struct otx2_nic *pfvf, struct otx2_cq_queue *cq, dma_addr_t *dma) { - if (unlikely(__otx2_alloc_rbuf(pfvf, cq->rbpool, dma))) + if (unlikely(__otx2_alloc_rbuf(pfvf, cq->rbpool, dma, + cq->cq_idx, cq->pool_ptrs - 1))) return -ENOMEM; return 0; } @@ -884,7 +893,7 @@ void otx2_sqb_flush(struct otx2_nic *pfvf) #define RQ_PASS_LVL_AURA (255 - ((95 * 256) / 100)) /* RED when 95% is full */ #define RQ_DROP_LVL_AURA (255 - ((99 * 256) / 100)) /* Drop when 99% is full */ -static int otx2_rq_init(struct otx2_nic *pfvf, u16 qidx, u16 lpb_aura) +int otx2_rq_init(struct otx2_nic *pfvf, u16 qidx, u16 lpb_aura) { struct otx2_qset *qset = &pfvf->qset; struct nix_aq_enq_req *aq; @@ -1028,6 +1037,10 @@ int otx2_sq_init(struct otx2_nic *pfvf, u16 qidx, u16 sqb_aura) sq->stats.bytes = 0; sq->stats.pkts = 0; + /* Attach XSK_BUFF_POOL to XDP queue */ + if (qidx > pfvf->hw.xdp_queues) + otx2_attach_xsk_buff(pfvf, sq, (qidx - pfvf->hw.xdp_queues)); + chan_offset = qidx % pfvf->hw.tx_chan_cnt; err = pfvf->hw_ops->sq_aq_init(pfvf, qidx, chan_offset, sqb_aura); @@ -1041,12 +1054,13 @@ int otx2_sq_init(struct otx2_nic *pfvf, u16 qidx, u16 sqb_aura) } -static int otx2_cq_init(struct otx2_nic *pfvf, u16 qidx) +int otx2_cq_init(struct otx2_nic *pfvf, u16 qidx) { struct otx2_qset *qset = &pfvf->qset; int err, pool_id, non_xdp_queues; struct nix_aq_enq_req *aq; struct otx2_cq_queue *cq; + struct otx2_pool *pool; cq = &qset->cq[qidx]; cq->cq_idx = qidx; @@ -1055,8 +1069,20 @@ static int otx2_cq_init(struct otx2_nic *pfvf, u16 qidx) cq->cq_type = CQ_RX; cq->cint_idx = qidx; cq->cqe_cnt = qset->rqe_cnt; - if (pfvf->xdp_prog) + if (pfvf->xdp_prog) { xdp_rxq_info_reg(&cq->xdp_rxq, pfvf->netdev, qidx, 0); + pool = &qset->pool[qidx]; + if (pool->xsk_pool) { + xdp_rxq_info_reg_mem_model(&cq->xdp_rxq, + MEM_TYPE_XSK_BUFF_POOL, + NULL); + xsk_pool_set_rxq_info(pool->xsk_pool, &cq->xdp_rxq); + } else if (pool->page_pool) { + xdp_rxq_info_reg_mem_model(&cq->xdp_rxq, + MEM_TYPE_PAGE_POOL, + pool->page_pool); + } + } } else if (qidx < non_xdp_queues) { cq->cq_type = CQ_TX; cq->cint_idx = qidx - pfvf->hw.rx_queues; @@ -1275,9 +1301,10 @@ void otx2_free_bufs(struct otx2_nic *pfvf, struct otx2_pool *pool, pa = otx2_iova_to_phys(pfvf->iommu_domain, iova); page = virt_to_head_page(phys_to_virt(pa)); - if (pool->page_pool) { page_pool_put_full_page(pool->page_pool, page, true); + } else if (pool->xsk_pool) { + /* Note: No way of identifying xdp_buff */ } else { dma_unmap_page_attrs(pfvf->dev, iova, size, DMA_FROM_DEVICE, @@ -1292,6 +1319,7 @@ void otx2_free_aura_ptr(struct otx2_nic *pfvf, int type) int pool_id, pool_start = 0, pool_end = 0, size = 0; struct otx2_pool *pool; u64 iova; + int idx; if (type == AURA_NIX_SQ) { pool_start = otx2_get_pool_idx(pfvf, type, 0); @@ -1306,16 +1334,21 @@ void otx2_free_aura_ptr(struct otx2_nic *pfvf, int type) /* Free SQB and RQB pointers from the aura pool */ for (pool_id = pool_start; pool_id < pool_end; pool_id++) { - iova = otx2_aura_allocptr(pfvf, pool_id); pool = &pfvf->qset.pool[pool_id]; + iova = otx2_aura_allocptr(pfvf, pool_id); while (iova) { if (type == AURA_NIX_RQ) iova -= OTX2_HEAD_ROOM; - otx2_free_bufs(pfvf, pool, iova, size); - iova = otx2_aura_allocptr(pfvf, pool_id); } + + for (idx = 0 ; idx < pool->xdp_cnt; idx++) { + if (!pool->xdp[idx]) + continue; + + xsk_buff_free(pool->xdp[idx]); + } } } @@ -1332,7 +1365,8 @@ void otx2_aura_pool_free(struct otx2_nic *pfvf) qmem_free(pfvf->dev, pool->stack); qmem_free(pfvf->dev, pool->fc_addr); page_pool_destroy(pool->page_pool); - pool->page_pool = NULL; + devm_kfree(pfvf->dev, pool->xdp); + pool->xsk_pool = NULL; } devm_kfree(pfvf->dev, pfvf->qset.pool); pfvf->qset.pool = NULL; @@ -1419,6 +1453,7 @@ int otx2_pool_init(struct otx2_nic *pfvf, u16 pool_id, int stack_pages, int numptrs, int buf_size, int type) { struct page_pool_params pp_params = { 0 }; + struct xsk_buff_pool *xsk_pool; struct npa_aq_enq_req *aq; struct otx2_pool *pool; int err; @@ -1462,21 +1497,35 @@ int otx2_pool_init(struct otx2_nic *pfvf, u16 pool_id, aq->ctype = NPA_AQ_CTYPE_POOL; aq->op = NPA_AQ_INSTOP_INIT; - if (type != AURA_NIX_RQ) { - pool->page_pool = NULL; + if (type != AURA_NIX_RQ) + return 0; + + if (!test_bit(pool_id, pfvf->af_xdp_zc_qidx)) { + pp_params.order = get_order(buf_size); + pp_params.flags = PP_FLAG_DMA_MAP; + pp_params.pool_size = min(OTX2_PAGE_POOL_SZ, numptrs); + pp_params.nid = NUMA_NO_NODE; + pp_params.dev = pfvf->dev; + pp_params.dma_dir = DMA_FROM_DEVICE; + pool->page_pool = page_pool_create(&pp_params); + if (IS_ERR(pool->page_pool)) { + netdev_err(pfvf->netdev, "Creation of page pool failed\n"); + return PTR_ERR(pool->page_pool); + } return 0; } - pp_params.order = get_order(buf_size); - pp_params.flags = PP_FLAG_DMA_MAP; - pp_params.pool_size = min(OTX2_PAGE_POOL_SZ, numptrs); - pp_params.nid = NUMA_NO_NODE; - pp_params.dev = pfvf->dev; - pp_params.dma_dir = DMA_FROM_DEVICE; - pool->page_pool = page_pool_create(&pp_params); - if (IS_ERR(pool->page_pool)) { - netdev_err(pfvf->netdev, "Creation of page pool failed\n"); - return PTR_ERR(pool->page_pool); + /* Set XSK pool to support AF_XDP zero-copy */ + xsk_pool = xsk_get_pool_from_qid(pfvf->netdev, pool_id); + if (xsk_pool) { + pool->xsk_pool = xsk_pool; + pool->xdp_cnt = numptrs; + pool->xdp = devm_kcalloc(pfvf->dev, + numptrs, sizeof(struct xdp_buff *), GFP_KERNEL); + if (IS_ERR(pool->xdp)) { + netdev_err(pfvf->netdev, "Creation of xsk pool failed\n"); + return PTR_ERR(pool->xdp); + } } return 0; @@ -1537,9 +1586,18 @@ int otx2_sq_aura_pool_init(struct otx2_nic *pfvf) } for (ptr = 0; ptr < num_sqbs; ptr++) { - err = otx2_alloc_rbuf(pfvf, pool, &bufptr); - if (err) + err = otx2_alloc_rbuf(pfvf, pool, &bufptr, pool_id, ptr); + if (err) { + if (pool->xsk_pool) { + ptr--; + while (ptr >= 0) { + xsk_buff_free(pool->xdp[ptr]); + ptr--; + } + } goto err_mem; + } + pfvf->hw_ops->aura_freeptr(pfvf, pool_id, bufptr); sq->sqb_ptrs[sq->sqb_count++] = (u64)bufptr; } @@ -1589,11 +1647,19 @@ int otx2_rq_aura_pool_init(struct otx2_nic *pfvf) /* Allocate pointers and free them to aura/pool */ for (pool_id = 0; pool_id < hw->rqpool_cnt; pool_id++) { pool = &pfvf->qset.pool[pool_id]; + for (ptr = 0; ptr < num_ptrs; ptr++) { - err = otx2_alloc_rbuf(pfvf, pool, &bufptr); - if (err) + err = otx2_alloc_rbuf(pfvf, pool, &bufptr, pool_id, ptr); + if (err) { + if (pool->xsk_pool) { + while (ptr) + xsk_buff_free(pool->xdp[--ptr]); + } return -ENOMEM; + } + pfvf->hw_ops->aura_freeptr(pfvf, pool_id, + pool->xsk_pool ? bufptr : bufptr + OTX2_HEAD_ROOM); } } @@ -1756,7 +1822,7 @@ int otx2_nix_config_bp(struct otx2_nic *pfvf, bool enable) req->chan_cnt = IEEE_8021QAZ_MAX_TCS; req->bpid_per_chan = 1; } else { - req->chan_cnt = 1; + req->chan_cnt = pfvf->hw.rx_chan_cnt; req->bpid_per_chan = 0; } @@ -1781,7 +1847,7 @@ int otx2_nix_cpt_config_bp(struct otx2_nic *pfvf, bool enable) req->chan_cnt = IEEE_8021QAZ_MAX_TCS; req->bpid_per_chan = 1; } else { - req->chan_cnt = 1; + req->chan_cnt = pfvf->hw.rx_chan_cnt; req->bpid_per_chan = 0; } @@ -1989,6 +2055,43 @@ int otx2_handle_ntuple_tc_features(struct net_device *netdev, netdev_features_t } EXPORT_SYMBOL(otx2_handle_ntuple_tc_features); +int otx2_set_hw_capabilities(struct otx2_nic *pfvf) +{ + struct mbox *mbox = &pfvf->mbox; + struct otx2_hw *hw = &pfvf->hw; + struct get_hw_cap_rsp *rsp; + struct msg_req *req; + int ret = -ENOMEM; + + mutex_lock(&mbox->lock); + + req = otx2_mbox_alloc_msg_get_hw_cap(mbox); + if (!req) + goto fail; + + ret = otx2_sync_mbox_msg(mbox); + if (ret) + goto fail; + + rsp = (struct get_hw_cap_rsp *)otx2_mbox_get_rsp(&pfvf->mbox.mbox, + 0, &req->hdr); + if (IS_ERR(rsp)) { + ret = -EINVAL; + goto fail; + } + + if (rsp->hw_caps & HW_CAP_MACSEC) + __set_bit(CN10K_HW_MACSEC, &hw->cap_flag); + + mutex_unlock(&mbox->lock); + + return 0; +fail: + dev_err(pfvf->dev, "Cannot get MACSEC capability from AF\n"); + mutex_unlock(&mbox->lock); + return ret; +} + #define M(_name, _id, _fn_name, _req_type, _rsp_type) \ int __weak \ otx2_mbox_up_handler_ ## _fn_name(struct otx2_nic *pfvf, \ diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h index 65814e3dc93f..ca0e6ab12ceb 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h @@ -21,6 +21,7 @@ #include <linux/time64.h> #include <linux/dim.h> #include <uapi/linux/if_macsec.h> +#include <net/page_pool/helpers.h> #include <mbox.h> #include <npc.h> @@ -128,6 +129,12 @@ enum otx2_errcodes_re { ERRCODE_IL4_CSUM = 0x22, }; +enum otx2_xdp_action { + OTX2_XDP_TX = BIT(0), + OTX2_XDP_REDIRECT = BIT(1), + OTX2_AF_XDP_FRAME = BIT(2), +}; + struct otx2_dev_stats { u64 rx_bytes; u64 rx_frames; @@ -349,6 +356,7 @@ struct otx2_flow_config { struct list_head flow_list_tc; u8 ucast_flt_cnt; bool ntuple; + u16 ntuple_cnt; }; struct dev_hw_ops { @@ -531,6 +539,8 @@ struct otx2_nic { /* Inline ipsec */ struct cn10k_ipsec ipsec; + /* af_xdp zero-copy */ + unsigned long *af_xdp_zc_qidx; }; static inline bool is_otx2_lbkvf(struct pci_dev *pdev) @@ -622,9 +632,6 @@ static inline void otx2_setup_dev_hw_settings(struct otx2_nic *pfvf) __set_bit(CN10K_PTP_ONESTEP, &hw->cap_flag); __set_bit(QOS_CIR_PIR_SUPPORT, &hw->cap_flag); } - - if (is_dev_cn10kb(pfvf->pdev)) - __set_bit(CN10K_HW_MACSEC, &hw->cap_flag); } /* Register read/write APIs */ @@ -862,6 +869,7 @@ static struct _req_type __maybe_unused \ *otx2_mbox_alloc_msg_ ## _fn_name(struct mbox *mbox) \ { \ struct _req_type *req; \ + u16 pcifunc = mbox->pfvf->pcifunc; \ \ req = (struct _req_type *)otx2_mbox_alloc_msg_rsp( \ &mbox->mbox, 0, sizeof(struct _req_type), \ @@ -870,7 +878,8 @@ static struct _req_type __maybe_unused \ return NULL; \ req->hdr.sig = OTX2_MBOX_REQ_SIG; \ req->hdr.id = _id; \ - trace_otx2_msg_alloc(mbox->mbox.pdev, _id, sizeof(*req)); \ + req->hdr.pcifunc = pcifunc; \ + trace_otx2_msg_alloc(mbox->mbox.pdev, _id, sizeof(*req), pcifunc); \ return req; \ } @@ -1002,7 +1011,7 @@ void otx2_txschq_free_one(struct otx2_nic *pfvf, u16 lvl, u16 schq); void otx2_free_pending_sqe(struct otx2_nic *pfvf); void otx2_sqb_flush(struct otx2_nic *pfvf); int otx2_alloc_rbuf(struct otx2_nic *pfvf, struct otx2_pool *pool, - dma_addr_t *dma); + dma_addr_t *dma, int qidx, int idx); int otx2_rxtx_enable(struct otx2_nic *pfvf, bool enable); void otx2_ctx_disable(struct mbox *mbox, int type, bool npa); int otx2_nix_config_bp(struct otx2_nic *pfvf, bool enable); @@ -1032,6 +1041,9 @@ void otx2_pfaf_mbox_destroy(struct otx2_nic *pf); void otx2_disable_mbox_intr(struct otx2_nic *pf); void otx2_disable_napi(struct otx2_nic *pf); irqreturn_t otx2_cq_intr_handler(int irq, void *cq_irq); +int otx2_rq_init(struct otx2_nic *pfvf, u16 qidx, u16 lpb_aura); +int otx2_cq_init(struct otx2_nic *pfvf, u16 qidx); +int otx2_set_hw_capabilities(struct otx2_nic *pfvf); /* RSS configuration APIs*/ int otx2_rss_init(struct otx2_nic *pfvf); @@ -1094,7 +1106,10 @@ int otx2_del_macfilter(struct net_device *netdev, const u8 *mac); int otx2_add_macfilter(struct net_device *netdev, const u8 *mac); int otx2_enable_rxvlan(struct otx2_nic *pf, bool enable); int otx2_install_rxvlan_offload_flow(struct otx2_nic *pfvf); -bool otx2_xdp_sq_append_pkt(struct otx2_nic *pfvf, u64 iova, int len, u16 qidx); +bool otx2_xdp_sq_append_pkt(struct otx2_nic *pfvf, struct xdp_frame *xdpf, + u64 iova, int len, u16 qidx, u16 flags); +void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq, struct xdp_frame *xdpf, + u64 dma_addr, int len, int *offset, u16 flags); u16 otx2_get_max_mtu(struct otx2_nic *pfvf); int otx2_handle_ntuple_tc_features(struct net_device *netdev, netdev_features_t features); @@ -1175,4 +1190,5 @@ static inline int mcam_entry_cmp(const void *a, const void *b) dma_addr_t otx2_dma_map_skb_frag(struct otx2_nic *pfvf, struct sk_buff *skb, int seg, int *len); void otx2_dma_unmap_skb_frags(struct otx2_nic *pfvf, struct sg_list *sg); +int otx2_read_free_sqe(struct otx2_nic *pfvf, u16 qidx); #endif /* OTX2_COMMON_H */ diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c index 33ec9a7f7c03..e13ae5484c19 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c @@ -41,6 +41,7 @@ static int otx2_dl_mcam_count_set(struct devlink *devlink, u32 id, if (!pfvf->flow_cfg) return 0; + pfvf->flow_cfg->ntuple_cnt = ctx->val.vu16; otx2_alloc_mcam_entries(pfvf, ctx->val.vu16); return 0; diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c index 2d53dc77ef1e..45b8c9230184 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c @@ -315,7 +315,7 @@ static void otx2_get_pauseparam(struct net_device *netdev, struct otx2_nic *pfvf = netdev_priv(netdev); struct cgx_pause_frm_cfg *req, *rsp; - if (is_otx2_lbkvf(pfvf->pdev)) + if (is_otx2_lbkvf(pfvf->pdev) || is_otx2_sdp_rep(pfvf->pdev)) return; mutex_lock(&pfvf->mbox.lock); @@ -347,7 +347,7 @@ static int otx2_set_pauseparam(struct net_device *netdev, if (pause->autoneg) return -EOPNOTSUPP; - if (is_otx2_lbkvf(pfvf->pdev)) + if (is_otx2_lbkvf(pfvf->pdev) || is_otx2_sdp_rep(pfvf->pdev)) return -EOPNOTSUPP; if (pause->rx_pause) @@ -910,8 +910,12 @@ static int otx2_get_rxfh(struct net_device *dev, return -ENOENT; if (indir) { - for (idx = 0; idx < rss->rss_size; idx++) + for (idx = 0; idx < rss->rss_size; idx++) { + /* Ignore if the rx queue is AF_XDP zero copy enabled */ + if (test_bit(rss_ctx->ind_tbl[idx], pfvf->af_xdp_zc_qidx)) + continue; indir[idx] = rss_ctx->ind_tbl[idx]; + } } if (rxfh->key) memcpy(rxfh->key, rss->key, sizeof(rss->key)); @@ -937,8 +941,8 @@ static u32 otx2_get_link(struct net_device *netdev) { struct otx2_nic *pfvf = netdev_priv(netdev); - /* LBK link is internal and always UP */ - if (is_otx2_lbkvf(pfvf->pdev)) + /* LBK and SDP links are internal and always UP */ + if (is_otx2_lbkvf(pfvf->pdev) || is_otx2_sdp_rep(pfvf->pdev)) return 1; return pfvf->linfo.link_up; } @@ -1409,7 +1413,7 @@ static int otx2vf_get_link_ksettings(struct net_device *netdev, { struct otx2_nic *pfvf = netdev_priv(netdev); - if (is_otx2_lbkvf(pfvf->pdev)) { + if (is_otx2_lbkvf(pfvf->pdev) || is_otx2_sdp_rep(pfvf->pdev)) { cmd->base.duplex = DUPLEX_FULL; cmd->base.speed = SPEED_100000; } else { diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c index 47bfd1fb37d4..64c6d9162ef6 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c @@ -247,7 +247,7 @@ int otx2_mcam_entry_init(struct otx2_nic *pfvf) mutex_unlock(&pfvf->mbox.lock); /* Allocate entries for Ntuple filters */ - count = otx2_alloc_mcam_entries(pfvf, OTX2_DEFAULT_FLOWCOUNT); + count = otx2_alloc_mcam_entries(pfvf, flow_cfg->ntuple_cnt); if (count <= 0) { otx2_clear_ntuple_flow_info(pfvf, flow_cfg); return 0; @@ -307,6 +307,7 @@ int otx2_mcam_flow_init(struct otx2_nic *pf) INIT_LIST_HEAD(&pf->flow_cfg->flow_list_tc); pf->flow_cfg->ucast_flt_cnt = OTX2_DEFAULT_UNICAST_FLOWS; + pf->flow_cfg->ntuple_cnt = OTX2_DEFAULT_FLOWCOUNT; /* Allocate bare minimum number of MCAM entries needed for * unicast and ntuple filters. diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c index e1dde93e8af8..db7c466fdc39 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c @@ -27,6 +27,7 @@ #include "qos.h" #include <rvu_trace.h> #include "cn10k_ipsec.h" +#include "otx2_xsk.h" #define DRV_NAME "rvu_nicpf" #define DRV_STRING "Marvell RVU NIC Physical Function Driver" @@ -464,6 +465,9 @@ static void otx2_pfvf_mbox_handler(struct work_struct *work) offset = ALIGN(sizeof(struct mbox_hdr), MBOX_MSG_ALIGN); + trace_otx2_msg_status(pf->pdev, "PF-VF down queue handler(forwarding)", + vf_mbox->num_msgs); + for (id = 0; id < vf_mbox->num_msgs; id++) { msg = (struct mbox_msghdr *)(mdev->mbase + mbox->rx_start + offset); @@ -472,7 +476,7 @@ static void otx2_pfvf_mbox_handler(struct work_struct *work) goto inval_msg; /* Set VF's number in each of the msg */ - msg->pcifunc &= RVU_PFVF_FUNC_MASK; + msg->pcifunc &= ~RVU_PFVF_FUNC_MASK; msg->pcifunc |= (vf_idx + 1) & RVU_PFVF_FUNC_MASK; offset = msg->next_msgoff; } @@ -502,6 +506,9 @@ static void otx2_pfvf_mbox_up_handler(struct work_struct *work) offset = mbox->rx_start + ALIGN(sizeof(struct mbox_hdr), MBOX_MSG_ALIGN); + trace_otx2_msg_status(pf->pdev, "PF-VF up queue handler(response)", + vf_mbox->up_num_msgs); + for (id = 0; id < vf_mbox->up_num_msgs; id++) { msg = mdev->mbase + offset; @@ -818,6 +825,9 @@ static void otx2_pfaf_mbox_handler(struct work_struct *work) offset = mbox->rx_start + ALIGN(sizeof(*rsp_hdr), MBOX_MSG_ALIGN); pf = af_mbox->pfvf; + trace_otx2_msg_status(pf->pdev, "PF-AF down queue handler(response)", + num_msgs); + for (id = 0; id < num_msgs; id++) { msg = (struct mbox_msghdr *)(mdev->mbase + offset); otx2_process_pfaf_mbox_msg(pf, msg); @@ -973,6 +983,9 @@ static void otx2_pfaf_mbox_up_handler(struct work_struct *work) offset = mbox->rx_start + ALIGN(sizeof(*rsp_hdr), MBOX_MSG_ALIGN); + trace_otx2_msg_status(pf->pdev, "PF-AF up queue handler(notification)", + num_msgs); + for (id = 0; id < num_msgs; id++) { msg = (struct mbox_msghdr *)(mdev->mbase + offset); @@ -1022,6 +1035,9 @@ static irqreturn_t otx2_pfaf_mbox_intr_handler(int irq, void *pf_irq) trace_otx2_msg_interrupt(pf->pdev, "UP message from AF to PF", BIT_ULL(0)); + + trace_otx2_msg_status(pf->pdev, "PF-AF up work queued(interrupt)", + hdr->num_msgs); } if (mbox_data & MBOX_DOWN_MSG) { @@ -1038,6 +1054,9 @@ static irqreturn_t otx2_pfaf_mbox_intr_handler(int irq, void *pf_irq) trace_otx2_msg_interrupt(pf->pdev, "DOWN reply from AF to PF", BIT_ULL(0)); + + trace_otx2_msg_status(pf->pdev, "PF-AF down work queued(interrupt)", + hdr->num_msgs); } return IRQ_HANDLED; @@ -1662,9 +1681,7 @@ void otx2_free_hw_resources(struct otx2_nic *pf) struct nix_lf_free_req *free_req; struct mbox *mbox = &pf->mbox; struct otx2_cq_queue *cq; - struct otx2_pool *pool; struct msg_req *req; - int pool_id; int qidx; /* Ensure all SQE are processed */ @@ -1705,13 +1722,6 @@ void otx2_free_hw_resources(struct otx2_nic *pf) /* Free RQ buffer pointers*/ otx2_free_aura_ptr(pf, AURA_NIX_RQ); - for (qidx = 0; qidx < pf->hw.rx_queues; qidx++) { - pool_id = otx2_get_pool_idx(pf, AURA_NIX_RQ, qidx); - pool = &pf->qset.pool[pool_id]; - page_pool_destroy(pool->page_pool); - pool->page_pool = NULL; - } - otx2_free_cq_res(pf); /* Free all ingress bandwidth profiles allocated */ @@ -2691,7 +2701,6 @@ static int otx2_get_vf_config(struct net_device *netdev, int vf, static int otx2_xdp_xmit_tx(struct otx2_nic *pf, struct xdp_frame *xdpf, int qidx) { - struct page *page; u64 dma_addr; int err = 0; @@ -2701,11 +2710,11 @@ static int otx2_xdp_xmit_tx(struct otx2_nic *pf, struct xdp_frame *xdpf, if (dma_mapping_error(pf->dev, dma_addr)) return -ENOMEM; - err = otx2_xdp_sq_append_pkt(pf, dma_addr, xdpf->len, qidx); + err = otx2_xdp_sq_append_pkt(pf, xdpf, dma_addr, xdpf->len, + qidx, OTX2_XDP_REDIRECT); if (!err) { otx2_dma_unmap_page(pf, dma_addr, xdpf->len, DMA_TO_DEVICE); - page = virt_to_page(xdpf->data); - put_page(page); + xdp_return_frame(xdpf); return -ENOMEM; } return 0; @@ -2789,6 +2798,8 @@ static int otx2_xdp(struct net_device *netdev, struct netdev_bpf *xdp) switch (xdp->command) { case XDP_SETUP_PROG: return otx2_xdp_setup(pf, xdp->prog); + case XDP_SETUP_XSK_POOL: + return otx2_xsk_pool_setup(pf, xdp->xsk.pool, xdp->xsk.queue_id); default: return -EINVAL; } @@ -2866,6 +2877,7 @@ static const struct net_device_ops otx2_netdev_ops = { .ndo_set_vf_vlan = otx2_set_vf_vlan, .ndo_get_vf_config = otx2_get_vf_config, .ndo_bpf = otx2_xdp, + .ndo_xsk_wakeup = otx2_xsk_wakeup, .ndo_xdp_xmit = otx2_xdp_xmit, .ndo_setup_tc = otx2_setup_tc, .ndo_set_vf_trust = otx2_ndo_set_vf_trust, @@ -3054,7 +3066,7 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id) return err; } - err = pci_request_regions(pdev, DRV_NAME); + err = pcim_request_all_regions(pdev, DRV_NAME); if (err) { dev_err(dev, "PCI request regions failed 0x%x\n", err); return err; @@ -3063,7 +3075,7 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id) err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(48)); if (err) { dev_err(dev, "DMA mask config failed, abort\n"); - goto err_release_regions; + return err; } pci_set_master(pdev); @@ -3073,10 +3085,8 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id) qos_txqs = min_t(int, qcount, OTX2_QOS_MAX_LEAF_NODES); netdev = alloc_etherdev_mqs(sizeof(*pf), qcount + qos_txqs, qcount); - if (!netdev) { - err = -ENOMEM; - goto err_release_regions; - } + if (!netdev) + return -ENOMEM; pci_set_drvdata(pdev, netdev); SET_NETDEV_DEV(netdev, &pdev->dev); @@ -3134,6 +3144,8 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id) if (err) goto err_ptp_destroy; + otx2_set_hw_capabilities(pf); + err = cn10k_mcs_init(pf); if (err) goto err_del_mcam_entries; @@ -3204,16 +3216,28 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id) /* Enable link notifications */ otx2_cgx_config_linkevents(pf, true); + pf->af_xdp_zc_qidx = bitmap_zalloc(qcount, GFP_KERNEL); + if (!pf->af_xdp_zc_qidx) { + err = -ENOMEM; + goto err_sriov_cleannup; + } + #ifdef CONFIG_DCB err = otx2_dcbnl_set_ops(netdev); if (err) - goto err_pf_sriov_init; + goto err_free_zc_bmap; #endif otx2_qos_init(pf, qos_txqs); return 0; +#ifdef CONFIG_DCB +err_free_zc_bmap: + bitmap_free(pf->af_xdp_zc_qidx); +#endif +err_sriov_cleannup: + otx2_sriov_vfcfg_cleanup(pf); err_pf_sriov_init: otx2_shutdown_tc(pf); err_mcam_flow_del: @@ -3240,8 +3264,6 @@ err_detach_rsrc: err_free_netdev: pci_set_drvdata(pdev, NULL); free_netdev(netdev); -err_release_regions: - pci_release_regions(pdev); return err; } @@ -3283,6 +3305,7 @@ static void otx2_vf_link_event_task(struct work_struct *work) req = (struct cgx_link_info_msg *)msghdr; req->hdr.id = MBOX_MSG_CGX_LINK_EVENT; req->hdr.sig = OTX2_MBOX_REQ_SIG; + req->hdr.pcifunc = pf->pcifunc; memcpy(&req->link_info, &pf->linfo, sizeof(req->link_info)); otx2_mbox_wait_for_zero(&pf->mbox_pfvf[0].mbox_up, vf_idx); @@ -3441,8 +3464,6 @@ static void otx2_remove(struct pci_dev *pdev) pci_free_irq_vectors(pf->pdev); pci_set_drvdata(pdev, NULL); free_netdev(netdev); - - pci_release_regions(pdev); } static struct pci_driver otx2_pf_driver = { diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c index 224cef938927..99ace381cc78 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c @@ -12,6 +12,7 @@ #include <linux/bpf_trace.h> #include <net/ip6_checksum.h> #include <net/xfrm.h> +#include <net/xdp.h> #include "otx2_reg.h" #include "otx2_common.h" @@ -19,6 +20,7 @@ #include "otx2_txrx.h" #include "otx2_ptp.h" #include "cn10k.h" +#include "otx2_xsk.h" #define CQE_ADDR(CQ, idx) ((CQ)->cqe_base + ((CQ)->cqe_size * (idx))) #define PTP_PORT 0x13F @@ -29,11 +31,17 @@ DEFINE_STATIC_KEY_FALSE(cn10k_ipsec_sa_enabled); +static int otx2_get_free_sqe(struct otx2_snd_queue *sq) +{ + return (sq->cons_head - sq->head - 1 + sq->sqe_cnt) + & (sq->sqe_cnt - 1); +} + static bool otx2_xdp_rcv_pkt_handler(struct otx2_nic *pfvf, struct bpf_prog *prog, struct nix_cqe_rx_s *cqe, struct otx2_cq_queue *cq, - bool *need_xdp_flush); + u32 *metasize, bool *need_xdp_flush); static void otx2_sq_set_sqe_base(struct otx2_snd_queue *sq, struct sk_buff *skb) @@ -96,20 +104,22 @@ static unsigned int frag_num(unsigned int i) static void otx2_xdp_snd_pkt_handler(struct otx2_nic *pfvf, struct otx2_snd_queue *sq, - struct nix_cqe_tx_s *cqe) + struct nix_cqe_tx_s *cqe, + int *xsk_frames) { struct nix_send_comp_s *snd_comp = &cqe->comp; struct sg_list *sg; - struct page *page; - u64 pa; sg = &sq->sg[snd_comp->sqe_id]; + if (sg->flags & OTX2_AF_XDP_FRAME) { + (*xsk_frames)++; + return; + } - pa = otx2_iova_to_phys(pfvf->iommu_domain, sg->dma_addr[0]); - otx2_dma_unmap_page(pfvf, sg->dma_addr[0], - sg->size[0], DMA_TO_DEVICE); - page = virt_to_page(phys_to_virt(pa)); - put_page(page); + if (sg->flags & OTX2_XDP_REDIRECT) + otx2_dma_unmap_page(pfvf, sg->dma_addr[0], sg->size[0], DMA_TO_DEVICE); + xdp_return_frame((struct xdp_frame *)sg->skb); + sg->skb = (u64)NULL; } static void otx2_snd_pkt_handler(struct otx2_nic *pfvf, @@ -325,18 +335,24 @@ static void otx2_rcv_pkt_handler(struct otx2_nic *pfvf, struct nix_rx_parse_s *parse = &cqe->parse; struct nix_rx_sg_s *sg = &cqe->sg; struct sk_buff *skb = NULL; + u64 *word = (u64 *)parse; void *end, *start; + u32 metasize = 0; u64 *seg_addr; u16 *seg_size; int seg; if (unlikely(parse->errlev || parse->errcode)) { - if (otx2_check_rcv_errors(pfvf, cqe, cq->cq_idx)) + if (otx2_check_rcv_errors(pfvf, cqe, cq->cq_idx)) { + trace_otx2_parse_dump(pfvf->pdev, "Err:", word); return; + } } + trace_otx2_parse_dump(pfvf->pdev, "", word); if (pfvf->xdp_prog) - if (otx2_xdp_rcv_pkt_handler(pfvf, pfvf->xdp_prog, cqe, cq, need_xdp_flush)) + if (otx2_xdp_rcv_pkt_handler(pfvf, pfvf->xdp_prog, cqe, cq, + &metasize, need_xdp_flush)) return; skb = napi_get_frags(napi); @@ -368,6 +384,8 @@ static void otx2_rcv_pkt_handler(struct otx2_nic *pfvf, skb->mark = parse->match_id; skb_mark_for_recycle(skb); + if (metasize) + skb_metadata_set(skb, metasize); napi_gro_frags(napi); } @@ -431,6 +449,18 @@ int otx2_refill_pool_ptrs(void *dev, struct otx2_cq_queue *cq) return cnt - cq->pool_ptrs; } +static void otx2_zc_submit_pkts(struct otx2_nic *pfvf, struct xsk_buff_pool *xsk_pool, + int *xsk_frames, int qidx, int budget) +{ + if (*xsk_frames) + xsk_tx_completed(xsk_pool, *xsk_frames); + + if (xsk_uses_need_wakeup(xsk_pool)) + xsk_set_tx_need_wakeup(xsk_pool); + + otx2_zc_napi_handler(pfvf, xsk_pool, qidx, budget); +} + static int otx2_tx_napi_handler(struct otx2_nic *pfvf, struct otx2_cq_queue *cq, int budget) { @@ -439,16 +469,22 @@ static int otx2_tx_napi_handler(struct otx2_nic *pfvf, struct nix_cqe_tx_s *cqe; struct net_device *ndev; int processed_cqe = 0; + int xsk_frames = 0; + + qidx = cq->cq_idx - pfvf->hw.rx_queues; + sq = &pfvf->qset.sq[qidx]; if (cq->pend_cqe >= budget) goto process_cqe; - if (otx2_nix_cq_op_status(pfvf, cq) || !cq->pend_cqe) + if (otx2_nix_cq_op_status(pfvf, cq) || !cq->pend_cqe) { + if (sq->xsk_pool) + otx2_zc_submit_pkts(pfvf, sq->xsk_pool, &xsk_frames, + qidx, budget); return 0; + } process_cqe: - qidx = cq->cq_idx - pfvf->hw.rx_queues; - sq = &pfvf->qset.sq[qidx]; while (likely(processed_cqe < budget) && cq->pend_cqe) { cqe = (struct nix_cqe_tx_s *)otx2_get_next_cqe(cq); @@ -458,10 +494,8 @@ process_cqe: break; } - qidx = cq->cq_idx - pfvf->hw.rx_queues; - if (cq->cq_type == CQ_XDP) - otx2_xdp_snd_pkt_handler(pfvf, sq, cqe); + otx2_xdp_snd_pkt_handler(pfvf, sq, cqe, &xsk_frames); else otx2_snd_pkt_handler(pfvf, cq, &pfvf->qset.sq[qidx], cqe, budget, &tx_pkts, &tx_bytes); @@ -502,6 +536,10 @@ process_cqe: netif_carrier_ok(ndev)) netif_tx_wake_queue(txq); } + + if (sq->xsk_pool) + otx2_zc_submit_pkts(pfvf, sq->xsk_pool, &xsk_frames, qidx, budget); + return 0; } @@ -527,9 +565,10 @@ static void otx2_adjust_adaptive_coalese(struct otx2_nic *pfvf, struct otx2_cq_p int otx2_napi_handler(struct napi_struct *napi, int budget) { struct otx2_cq_queue *rx_cq = NULL; + struct otx2_cq_queue *cq = NULL; + struct otx2_pool *pool = NULL; struct otx2_cq_poll *cq_poll; int workdone = 0, cq_idx, i; - struct otx2_cq_queue *cq; struct otx2_qset *qset; struct otx2_nic *pfvf; int filled_cnt = -1; @@ -554,6 +593,7 @@ int otx2_napi_handler(struct napi_struct *napi, int budget) if (rx_cq && rx_cq->pool_ptrs) filled_cnt = pfvf->hw_ops->refill_pool_ptrs(pfvf, rx_cq); + /* Clear the IRQ */ otx2_write64(pfvf, NIX_LF_CINTX_INT(cq_poll->cint_idx), BIT_ULL(0)); @@ -566,20 +606,31 @@ int otx2_napi_handler(struct napi_struct *napi, int budget) if (pfvf->flags & OTX2_FLAG_ADPTV_INT_COAL_ENABLED) otx2_adjust_adaptive_coalese(pfvf, cq_poll); + if (likely(cq)) + pool = &pfvf->qset.pool[cq->cq_idx]; + if (unlikely(!filled_cnt)) { struct refill_work *work; struct delayed_work *dwork; - work = &pfvf->refill_wrk[cq->cq_idx]; - dwork = &work->pool_refill_work; - /* Schedule a task if no other task is running */ - if (!cq->refill_task_sched) { - work->napi = napi; - cq->refill_task_sched = true; - schedule_delayed_work(dwork, - msecs_to_jiffies(100)); + if (likely(cq)) { + work = &pfvf->refill_wrk[cq->cq_idx]; + dwork = &work->pool_refill_work; + /* Schedule a task if no other task is running */ + if (!cq->refill_task_sched) { + work->napi = napi; + cq->refill_task_sched = true; + schedule_delayed_work(dwork, + msecs_to_jiffies(100)); + } + /* Call wake-up for not able to fill buffers */ + if (pool->xsk_pool) + xsk_set_rx_need_wakeup(pool->xsk_pool); } } else { + /* Clear wake-up, since buffers are filled successfully */ + if (pool && pool->xsk_pool) + xsk_clear_rx_need_wakeup(pool->xsk_pool); /* Re-enable interrupts */ otx2_write64(pfvf, NIX_LF_CINTX_ENA_W1S(cq_poll->cint_idx), @@ -1147,7 +1198,7 @@ bool otx2_sq_append_skb(void *dev, struct netdev_queue *txq, /* Check if there is enough room between producer * and consumer index. */ - free_desc = (sq->cons_head - sq->head - 1 + sq->sqe_cnt) & (sq->sqe_cnt - 1); + free_desc = otx2_get_free_sqe(sq); if (free_desc < sq->sqe_thresh) return false; @@ -1230,15 +1281,19 @@ void otx2_cleanup_rx_cqes(struct otx2_nic *pfvf, struct otx2_cq_queue *cq, int q u16 pool_id; u64 iova; - if (pfvf->xdp_prog) + pool_id = otx2_get_pool_idx(pfvf, AURA_NIX_RQ, qidx); + pool = &pfvf->qset.pool[pool_id]; + + if (pfvf->xdp_prog) { + if (pool->page_pool) + xdp_rxq_info_unreg_mem_model(&cq->xdp_rxq); + xdp_rxq_info_unreg(&cq->xdp_rxq); + } if (otx2_nix_cq_op_status(pfvf, cq) || !cq->pend_cqe) return; - pool_id = otx2_get_pool_idx(pfvf, AURA_NIX_RQ, qidx); - pool = &pfvf->qset.pool[pool_id]; - while (cq->pend_cqe) { cqe = (struct nix_cqe_rx_s *)otx2_get_next_cqe(cq); processed_cqe++; @@ -1359,8 +1414,8 @@ void otx2_free_pending_sqe(struct otx2_nic *pfvf) } } -static void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq, u64 dma_addr, - int len, int *offset) +void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq, struct xdp_frame *xdpf, + u64 dma_addr, int len, int *offset, u16 flags) { struct nix_sqe_sg_s *sg = NULL; u64 *iova = NULL; @@ -1377,16 +1432,34 @@ static void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq, u64 dma_addr, sq->sg[sq->head].dma_addr[0] = dma_addr; sq->sg[sq->head].size[0] = len; sq->sg[sq->head].num_segs = 1; + sq->sg[sq->head].flags = flags; + sq->sg[sq->head].skb = (u64)xdpf; +} + +int otx2_read_free_sqe(struct otx2_nic *pfvf, u16 qidx) +{ + struct otx2_snd_queue *sq; + int free_sqe; + + sq = &pfvf->qset.sq[qidx]; + free_sqe = otx2_get_free_sqe(sq); + if (free_sqe < sq->sqe_thresh) { + netdev_warn(pfvf->netdev, "No free sqe for Send queue%d\n", qidx); + return 0; + } + + return free_sqe - sq->sqe_thresh; } -bool otx2_xdp_sq_append_pkt(struct otx2_nic *pfvf, u64 iova, int len, u16 qidx) +bool otx2_xdp_sq_append_pkt(struct otx2_nic *pfvf, struct xdp_frame *xdpf, + u64 iova, int len, u16 qidx, u16 flags) { struct nix_sqe_hdr_s *sqe_hdr; struct otx2_snd_queue *sq; int offset, free_sqe; sq = &pfvf->qset.sq[qidx]; - free_sqe = (sq->num_sqbs - *sq->aura_fc_addr) * sq->sqe_per_sqb; + free_sqe = otx2_get_free_sqe(sq); if (free_sqe < sq->sqe_thresh) return false; @@ -1405,7 +1478,7 @@ bool otx2_xdp_sq_append_pkt(struct otx2_nic *pfvf, u64 iova, int len, u16 qidx) offset = sizeof(*sqe_hdr); - otx2_xdp_sqe_add_sg(sq, iova, len, &offset); + otx2_xdp_sqe_add_sg(sq, xdpf, iova, len, &offset, flags); sqe_hdr->sizem1 = (offset / 16) - 1; pfvf->hw_ops->sqe_flush(pfvf, sq, offset, qidx); @@ -1416,16 +1489,30 @@ static bool otx2_xdp_rcv_pkt_handler(struct otx2_nic *pfvf, struct bpf_prog *prog, struct nix_cqe_rx_s *cqe, struct otx2_cq_queue *cq, - bool *need_xdp_flush) + u32 *metasize, bool *need_xdp_flush) { + struct xdp_buff xdp, *xsk_buff = NULL; unsigned char *hard_start; + struct otx2_pool *pool; + struct xdp_frame *xdpf; int qidx = cq->cq_idx; - struct xdp_buff xdp; struct page *page; u64 iova, pa; u32 act; int err; + pool = &pfvf->qset.pool[qidx]; + + if (pool->xsk_pool) { + xsk_buff = pool->xdp[--cq->rbpool->xdp_top]; + if (!xsk_buff) + return false; + + xsk_buff->data_end = xsk_buff->data + cqe->sg.seg_size; + act = bpf_prog_run_xdp(prog, xsk_buff); + goto handle_xdp_verdict; + } + iova = cqe->sg.seg_addr - OTX2_HEAD_ROOM; pa = otx2_iova_to_phys(pfvf->iommu_domain, iova); page = virt_to_page(phys_to_virt(pa)); @@ -1434,41 +1521,63 @@ static bool otx2_xdp_rcv_pkt_handler(struct otx2_nic *pfvf, hard_start = (unsigned char *)phys_to_virt(pa); xdp_prepare_buff(&xdp, hard_start, OTX2_HEAD_ROOM, - cqe->sg.seg_size, false); + cqe->sg.seg_size, true); act = bpf_prog_run_xdp(prog, &xdp); +handle_xdp_verdict: switch (act) { case XDP_PASS: + *metasize = xdp.data - xdp.data_meta; break; case XDP_TX: qidx += pfvf->hw.tx_queues; cq->pool_ptrs++; - return otx2_xdp_sq_append_pkt(pfvf, iova, - cqe->sg.seg_size, qidx); + xdpf = xdp_convert_buff_to_frame(&xdp); + return otx2_xdp_sq_append_pkt(pfvf, xdpf, + cqe->sg.seg_addr, + cqe->sg.seg_size, + qidx, OTX2_XDP_TX); case XDP_REDIRECT: cq->pool_ptrs++; - err = xdp_do_redirect(pfvf->netdev, &xdp, prog); + if (xsk_buff) { + err = xdp_do_redirect(pfvf->netdev, xsk_buff, prog); + if (!err) { + *need_xdp_flush = true; + return true; + } + return false; + } - otx2_dma_unmap_page(pfvf, iova, pfvf->rbsize, - DMA_FROM_DEVICE); + err = xdp_do_redirect(pfvf->netdev, &xdp, prog); if (!err) { *need_xdp_flush = true; return true; } - put_page(page); + + otx2_dma_unmap_page(pfvf, iova, pfvf->rbsize, + DMA_FROM_DEVICE); + xdpf = xdp_convert_buff_to_frame(&xdp); + xdp_return_frame(xdpf); break; default: bpf_warn_invalid_xdp_action(pfvf->netdev, prog, act); - break; + fallthrough; case XDP_ABORTED: - trace_xdp_exception(pfvf->netdev, prog, act); - break; + if (act == XDP_ABORTED) + trace_xdp_exception(pfvf->netdev, prog, act); + fallthrough; case XDP_DROP: - otx2_dma_unmap_page(pfvf, iova, pfvf->rbsize, - DMA_FROM_DEVICE); - put_page(page); cq->pool_ptrs++; + if (xsk_buff) { + xsk_buff_free(xsk_buff); + } else if (page->pp) { + page_pool_recycle_direct(pool->page_pool, page); + } else { + otx2_dma_unmap_page(pfvf, iova, pfvf->rbsize, + DMA_FROM_DEVICE); + put_page(page); + } return true; } return false; diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.h index d23810963fdb..acf259d72008 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.h +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.h @@ -12,6 +12,7 @@ #include <linux/iommu.h> #include <linux/if_vlan.h> #include <net/xdp.h> +#include <net/xdp_sock_drv.h> #define LBK_CHAN_BASE 0x000 #define SDP_CHAN_BASE 0x700 @@ -76,6 +77,7 @@ struct otx2_rcv_queue { struct sg_list { u16 num_segs; + u16 flags; u64 skb; u64 size[OTX2_MAX_FRAGS_IN_SQE]; u64 dma_addr[OTX2_MAX_FRAGS_IN_SQE]; @@ -104,6 +106,8 @@ struct otx2_snd_queue { /* SQE ring and CPT response queue for Inline IPSEC */ struct qmem *sqe_ring; struct qmem *cpt_resp; + /* Buffer pool for af_xdp zero-copy */ + struct xsk_buff_pool *xsk_pool; } ____cacheline_aligned_in_smp; enum cq_type { @@ -127,7 +131,11 @@ struct otx2_pool { struct qmem *stack; struct qmem *fc_addr; struct page_pool *page_pool; + struct xsk_buff_pool *xsk_pool; + struct xdp_buff **xdp; + u16 xdp_cnt; u16 rbsize; + u16 xdp_top; }; struct otx2_cq_queue { @@ -144,6 +152,7 @@ struct otx2_cq_queue { void *cqe_base; struct qmem *cqe; struct otx2_pool *rbpool; + bool xsk_zc_en; struct xdp_rxq_info xdp_rxq; } ____cacheline_aligned_in_smp; diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c index e926c6ce96cf..8a8b598bd389 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c @@ -136,7 +136,7 @@ static int otx2vf_process_mbox_msg_up(struct otx2_nic *vf, rsp->hdr.id = MBOX_MSG_CGX_LINK_EVENT; rsp->hdr.sig = OTX2_MBOX_RSP_SIG; - rsp->hdr.pcifunc = 0; + rsp->hdr.pcifunc = req->pcifunc; rsp->hdr.rc = 0; err = otx2_mbox_up_handler_cgx_link_event( vf, (struct cgx_link_info_msg *)req, rsp); @@ -548,7 +548,7 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id) return err; } - err = pci_request_regions(pdev, DRV_NAME); + err = pcim_request_all_regions(pdev, DRV_NAME); if (err) { dev_err(dev, "PCI request regions failed 0x%x\n", err); return err; @@ -557,7 +557,7 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id) err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(48)); if (err) { dev_err(dev, "DMA mask config failed, abort\n"); - goto err_release_regions; + return err; } pci_set_master(pdev); @@ -565,10 +565,8 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id) qcount = num_online_cpus(); qos_txqs = min_t(int, qcount, OTX2_QOS_MAX_LEAF_NODES); netdev = alloc_etherdev_mqs(sizeof(*vf), qcount + qos_txqs, qcount); - if (!netdev) { - err = -ENOMEM; - goto err_release_regions; - } + if (!netdev) + return -ENOMEM; pci_set_drvdata(pdev, netdev); SET_NETDEV_DEV(netdev, &pdev->dev); @@ -722,15 +720,30 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id) if (err) goto err_shutdown_tc; + vf->af_xdp_zc_qidx = bitmap_zalloc(qcount, GFP_KERNEL); + if (!vf->af_xdp_zc_qidx) { + err = -ENOMEM; + goto err_unreg_devlink; + } + #ifdef CONFIG_DCB - err = otx2_dcbnl_set_ops(netdev); - if (err) - goto err_shutdown_tc; + /* Priority flow control is not supported for LBK and SDP vf(s) */ + if (!(is_otx2_lbkvf(vf->pdev) || is_otx2_sdp_rep(vf->pdev))) { + err = otx2_dcbnl_set_ops(netdev); + if (err) + goto err_free_zc_bmap; + } #endif otx2_qos_init(vf, qos_txqs); return 0; +#ifdef CONFIG_DCB +err_free_zc_bmap: + bitmap_free(vf->af_xdp_zc_qidx); +#endif +err_unreg_devlink: + otx2_unregister_dl(vf); err_shutdown_tc: otx2_shutdown_tc(vf); err_unreg_netdev: @@ -753,8 +766,6 @@ err_free_irq_vectors: err_free_netdev: pci_set_drvdata(pdev, NULL); free_netdev(netdev); -err_release_regions: - pci_release_regions(pdev); return err; } @@ -803,8 +814,6 @@ static void otx2vf_remove(struct pci_dev *pdev) pci_free_irq_vectors(vf->pdev); pci_set_drvdata(pdev, NULL); free_netdev(netdev); - - pci_release_regions(pdev); } static struct pci_driver otx2vf_driver = { diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c new file mode 100644 index 000000000000..b328aae23d73 --- /dev/null +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c @@ -0,0 +1,245 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Marvell RVU Ethernet driver + * + * Copyright (C) 2024 Marvell. + * + */ + +#include <linux/bpf_trace.h> +#include <linux/stringify.h> +#include <net/xdp_sock_drv.h> +#include <net/xdp.h> + +#include "otx2_common.h" +#include "otx2_struct.h" +#include "otx2_xsk.h" + +int otx2_xsk_pool_alloc_buf(struct otx2_nic *pfvf, struct otx2_pool *pool, + dma_addr_t *dma, int idx) +{ + struct xdp_buff *xdp; + int delta; + + xdp = xsk_buff_alloc(pool->xsk_pool); + if (!xdp) + return -ENOMEM; + + pool->xdp[pool->xdp_top++] = xdp; + *dma = OTX2_DATA_ALIGN(xsk_buff_xdp_get_dma(xdp)); + /* Adjust xdp->data for unaligned addresses */ + delta = *dma - xsk_buff_xdp_get_dma(xdp); + xdp->data += delta; + + return 0; +} + +static int otx2_xsk_ctx_disable(struct otx2_nic *pfvf, u16 qidx, int aura_id) +{ + struct nix_cn10k_aq_enq_req *cn10k_rq_aq; + struct npa_aq_enq_req *aura_aq; + struct npa_aq_enq_req *pool_aq; + struct nix_aq_enq_req *rq_aq; + + if (test_bit(CN10K_LMTST, &pfvf->hw.cap_flag)) { + cn10k_rq_aq = otx2_mbox_alloc_msg_nix_cn10k_aq_enq(&pfvf->mbox); + if (!cn10k_rq_aq) + return -ENOMEM; + cn10k_rq_aq->qidx = qidx; + cn10k_rq_aq->rq.ena = 0; + cn10k_rq_aq->rq_mask.ena = 1; + cn10k_rq_aq->ctype = NIX_AQ_CTYPE_RQ; + cn10k_rq_aq->op = NIX_AQ_INSTOP_WRITE; + } else { + rq_aq = otx2_mbox_alloc_msg_nix_aq_enq(&pfvf->mbox); + if (!rq_aq) + return -ENOMEM; + rq_aq->qidx = qidx; + rq_aq->sq.ena = 0; + rq_aq->sq_mask.ena = 1; + rq_aq->ctype = NIX_AQ_CTYPE_RQ; + rq_aq->op = NIX_AQ_INSTOP_WRITE; + } + + aura_aq = otx2_mbox_alloc_msg_npa_aq_enq(&pfvf->mbox); + if (!aura_aq) + goto fail; + + aura_aq->aura_id = aura_id; + aura_aq->aura.ena = 0; + aura_aq->aura_mask.ena = 1; + aura_aq->ctype = NPA_AQ_CTYPE_AURA; + aura_aq->op = NPA_AQ_INSTOP_WRITE; + + pool_aq = otx2_mbox_alloc_msg_npa_aq_enq(&pfvf->mbox); + if (!pool_aq) + goto fail; + + pool_aq->aura_id = aura_id; + pool_aq->pool.ena = 0; + pool_aq->pool_mask.ena = 1; + + pool_aq->ctype = NPA_AQ_CTYPE_POOL; + pool_aq->op = NPA_AQ_INSTOP_WRITE; + + return otx2_sync_mbox_msg(&pfvf->mbox); + +fail: + otx2_mbox_reset(&pfvf->mbox.mbox, 0); + return -ENOMEM; +} + +static void otx2_clean_up_rq(struct otx2_nic *pfvf, int qidx) +{ + struct otx2_qset *qset = &pfvf->qset; + struct otx2_cq_queue *cq; + struct otx2_pool *pool; + u64 iova; + + /* If the DOWN flag is set SQs are already freed */ + if (pfvf->flags & OTX2_FLAG_INTF_DOWN) + return; + + cq = &qset->cq[qidx]; + if (cq) + otx2_cleanup_rx_cqes(pfvf, cq, qidx); + + pool = &pfvf->qset.pool[qidx]; + iova = otx2_aura_allocptr(pfvf, qidx); + while (iova) { + iova -= OTX2_HEAD_ROOM; + otx2_free_bufs(pfvf, pool, iova, pfvf->rbsize); + iova = otx2_aura_allocptr(pfvf, qidx); + } + + mutex_lock(&pfvf->mbox.lock); + otx2_xsk_ctx_disable(pfvf, qidx, qidx); + mutex_unlock(&pfvf->mbox.lock); +} + +int otx2_xsk_pool_enable(struct otx2_nic *pf, struct xsk_buff_pool *pool, u16 qidx) +{ + u16 rx_queues = pf->hw.rx_queues; + u16 tx_queues = pf->hw.tx_queues; + int err; + + if (qidx >= rx_queues || qidx >= tx_queues) + return -EINVAL; + + err = xsk_pool_dma_map(pool, pf->dev, DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING); + if (err) + return err; + + set_bit(qidx, pf->af_xdp_zc_qidx); + otx2_clean_up_rq(pf, qidx); + /* Reconfigure RSS table as 'qidx' cannot be part of RSS now */ + otx2_set_rss_table(pf, DEFAULT_RSS_CONTEXT_GROUP); + /* Kick start the NAPI context so that receiving will start */ + return otx2_xsk_wakeup(pf->netdev, qidx, XDP_WAKEUP_RX); +} + +int otx2_xsk_pool_disable(struct otx2_nic *pf, u16 qidx) +{ + struct net_device *netdev = pf->netdev; + struct xsk_buff_pool *pool; + struct otx2_snd_queue *sq; + + pool = xsk_get_pool_from_qid(netdev, qidx); + if (!pool) + return -EINVAL; + + sq = &pf->qset.sq[qidx + pf->hw.tx_queues]; + sq->xsk_pool = NULL; + otx2_clean_up_rq(pf, qidx); + clear_bit(qidx, pf->af_xdp_zc_qidx); + xsk_pool_dma_unmap(pool, DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING); + /* Reconfigure RSS table as 'qidx' now need to be part of RSS now */ + otx2_set_rss_table(pf, DEFAULT_RSS_CONTEXT_GROUP); + + return 0; +} + +int otx2_xsk_pool_setup(struct otx2_nic *pf, struct xsk_buff_pool *pool, u16 qidx) +{ + if (pool) + return otx2_xsk_pool_enable(pf, pool, qidx); + + return otx2_xsk_pool_disable(pf, qidx); +} + +int otx2_xsk_wakeup(struct net_device *dev, u32 queue_id, u32 flags) +{ + struct otx2_nic *pf = netdev_priv(dev); + struct otx2_cq_poll *cq_poll = NULL; + struct otx2_qset *qset = &pf->qset; + + if (pf->flags & OTX2_FLAG_INTF_DOWN) + return -ENETDOWN; + + if (queue_id >= pf->hw.rx_queues || queue_id >= pf->hw.tx_queues) + return -EINVAL; + + cq_poll = &qset->napi[queue_id]; + if (!cq_poll) + return -EINVAL; + + /* Trigger interrupt */ + if (!napi_if_scheduled_mark_missed(&cq_poll->napi)) { + otx2_write64(pf, NIX_LF_CINTX_ENA_W1S(cq_poll->cint_idx), BIT_ULL(0)); + otx2_write64(pf, NIX_LF_CINTX_INT_W1S(cq_poll->cint_idx), BIT_ULL(0)); + } + + return 0; +} + +void otx2_attach_xsk_buff(struct otx2_nic *pfvf, struct otx2_snd_queue *sq, int qidx) +{ + if (test_bit(qidx, pfvf->af_xdp_zc_qidx)) + sq->xsk_pool = xsk_get_pool_from_qid(pfvf->netdev, qidx); +} + +static void otx2_xsk_sq_append_pkt(struct otx2_nic *pfvf, u64 iova, int len, + u16 qidx) +{ + struct nix_sqe_hdr_s *sqe_hdr; + struct otx2_snd_queue *sq; + int offset; + + sq = &pfvf->qset.sq[qidx]; + memset(sq->sqe_base + 8, 0, sq->sqe_size - 8); + + sqe_hdr = (struct nix_sqe_hdr_s *)(sq->sqe_base); + + if (!sqe_hdr->total) { + sqe_hdr->aura = sq->aura_id; + sqe_hdr->df = 1; + sqe_hdr->sq = qidx; + sqe_hdr->pnc = 1; + } + sqe_hdr->total = len; + sqe_hdr->sqe_id = sq->head; + + offset = sizeof(*sqe_hdr); + + otx2_xdp_sqe_add_sg(sq, NULL, iova, len, &offset, OTX2_AF_XDP_FRAME); + sqe_hdr->sizem1 = (offset / 16) - 1; + pfvf->hw_ops->sqe_flush(pfvf, sq, offset, qidx); +} + +void otx2_zc_napi_handler(struct otx2_nic *pfvf, struct xsk_buff_pool *pool, + int queue, int budget) +{ + struct xdp_desc *xdp_desc = pool->tx_descs; + int i, batch; + + budget = min(budget, otx2_read_free_sqe(pfvf, queue)); + batch = xsk_tx_peek_release_desc_batch(pool, budget); + if (!batch) + return; + + for (i = 0; i < batch; i++) { + dma_addr_t dma_addr; + + dma_addr = xsk_buff_raw_get_dma(pool, xdp_desc[i].addr); + otx2_xsk_sq_append_pkt(pfvf, dma_addr, xdp_desc[i].len, queue); + } +} diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h new file mode 100644 index 000000000000..8047fafee8fe --- /dev/null +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h @@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Marvell RVU PF/VF Netdev Devlink + * + * Copyright (C) 2024 Marvell. + * + */ + +#ifndef OTX2_XSK_H +#define OTX2_XSK_H + +struct otx2_nic; +struct xsk_buff_pool; + +int otx2_xsk_pool_setup(struct otx2_nic *pf, struct xsk_buff_pool *pool, u16 qid); +int otx2_xsk_pool_enable(struct otx2_nic *pf, struct xsk_buff_pool *pool, u16 qid); +int otx2_xsk_pool_disable(struct otx2_nic *pf, u16 qid); +int otx2_xsk_pool_alloc_buf(struct otx2_nic *pfvf, struct otx2_pool *pool, + dma_addr_t *dma, int idx); +int otx2_xsk_wakeup(struct net_device *dev, u32 queue_id, u32 flags); +void otx2_zc_napi_handler(struct otx2_nic *pfvf, struct xsk_buff_pool *pool, + int queue, int budget); +void otx2_attach_xsk_buff(struct otx2_nic *pfvf, struct otx2_snd_queue *sq, int qidx); + +#endif /* OTX2_XSK_H */ diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/qos.c b/drivers/net/ethernet/marvell/octeontx2/nic/qos.c index 0f844c14485a..5765bac119f0 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/qos.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/qos.c @@ -165,6 +165,11 @@ static void __otx2_qos_txschq_cfg(struct otx2_nic *pfvf, otx2_config_sched_shaping(pfvf, node, cfg, &num_regs); } else if (level == NIX_TXSCH_LVL_TL2) { + /* configure parent txschq */ + cfg->reg[num_regs] = NIX_AF_TL2X_PARENT(node->schq); + cfg->regval[num_regs] = (u64)hw->tx_link << 16; + num_regs++; + /* configure link cfg */ if (level == pfvf->qos.link_cfg_lvl) { cfg->reg[num_regs] = NIX_AF_TL3_TL2X_LINKX_CFG(node->schq, hw->tx_link); @@ -1633,6 +1638,7 @@ static int otx2_qos_leaf_del_last(struct otx2_nic *pfvf, u16 classid, bool force if (!node->is_static) dwrr_del_node = true; + WRITE_ONCE(node->qid, OTX2_QOS_QID_INNER); /* destroy the leaf node */ otx2_qos_disable_sq(pfvf, qid); otx2_qos_destroy_node(pfvf, node); @@ -1677,9 +1683,6 @@ static int otx2_qos_leaf_del_last(struct otx2_nic *pfvf, u16 classid, bool force } kfree(new_cfg); - /* update tx_real_queues */ - otx2_qos_update_tx_netdev_queues(pfvf); - return 0; } diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c b/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c index 9d887bfc3108..58d572ce08ef 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c @@ -82,7 +82,7 @@ static int otx2_qos_sq_aura_pool_init(struct otx2_nic *pfvf, int qidx) } for (ptr = 0; ptr < num_sqbs; ptr++) { - err = otx2_alloc_rbuf(pfvf, pool, &bufptr); + err = otx2_alloc_rbuf(pfvf, pool, &bufptr, pool_id, ptr); if (err) goto sqb_free; pfvf->hw_ops->aura_freeptr(pfvf, pool_id, bufptr); @@ -256,6 +256,26 @@ out: return err; } +static int otx2_qos_nix_npa_ndc_sync(struct otx2_nic *pfvf) +{ + struct ndc_sync_op *req; + int rc; + + mutex_lock(&pfvf->mbox.lock); + + req = otx2_mbox_alloc_msg_ndc_sync_op(&pfvf->mbox); + if (!req) { + mutex_unlock(&pfvf->mbox.lock); + return -ENOMEM; + } + + req->nix_lf_tx_sync = true; + req->npa_lf_sync = true; + rc = otx2_sync_mbox_msg(&pfvf->mbox); + mutex_unlock(&pfvf->mbox.lock); + return rc; +} + void otx2_qos_disable_sq(struct otx2_nic *pfvf, int qidx) { struct otx2_qset *qset = &pfvf->qset; @@ -285,6 +305,8 @@ void otx2_qos_disable_sq(struct otx2_nic *pfvf, int qidx) otx2_qos_sqb_flush(pfvf, sq_idx); otx2_smq_flush(pfvf, otx2_get_smq_idx(pfvf, sq_idx)); + /* NIX/NPA NDC sync */ + otx2_qos_nix_npa_ndc_sync(pfvf); otx2_cleanup_tx_cqes(pfvf, cq); mutex_lock(&pfvf->mbox.lock); diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/rep.c b/drivers/net/ethernet/marvell/octeontx2/nic/rep.c index 04e08e06f30f..2cd3da3b6843 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/rep.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/rep.c @@ -67,6 +67,8 @@ static int rvu_rep_mcam_flow_init(struct rep_dev *rep) rsp = (struct npc_mcam_alloc_entry_rsp *)otx2_mbox_get_rsp (&priv->mbox.mbox, 0, &req->hdr); + if (IS_ERR(rsp)) + goto exit; for (ent = 0; ent < rsp->count; ent++) rep->flow_cfg->flow_ent[ent + allocated] = rsp->entry_list[ent]; @@ -763,7 +765,7 @@ static int rvu_rep_probe(struct pci_dev *pdev, const struct pci_device_id *id) return err; } - err = pci_request_regions(pdev, DRV_NAME); + err = pcim_request_all_regions(pdev, DRV_NAME); if (err) { dev_err(dev, "PCI request regions failed 0x%x\n", err); return err; @@ -772,7 +774,7 @@ static int rvu_rep_probe(struct pci_dev *pdev, const struct pci_device_id *id) err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(48)); if (err) { dev_err(dev, "DMA mask config failed, abort\n"); - goto err_release_regions; + goto err_set_drv_data; } pci_set_master(pdev); @@ -780,7 +782,7 @@ static int rvu_rep_probe(struct pci_dev *pdev, const struct pci_device_id *id) priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL); if (!priv) { err = -ENOMEM; - goto err_release_regions; + goto err_set_drv_data; } pci_set_drvdata(pdev, priv); @@ -797,7 +799,7 @@ static int rvu_rep_probe(struct pci_dev *pdev, const struct pci_device_id *id) err = otx2_init_rsrc(pdev, priv); if (err) - goto err_release_regions; + goto err_set_drv_data; priv->iommu_domain = iommu_get_domain_for_dev(dev); @@ -820,9 +822,8 @@ err_detach_rsrc: otx2_disable_mbox_intr(priv); otx2_pfaf_mbox_destroy(priv); pci_free_irq_vectors(pdev); -err_release_regions: +err_set_drv_data: pci_set_drvdata(pdev, NULL); - pci_release_regions(pdev); return err; } @@ -842,7 +843,6 @@ static void rvu_rep_remove(struct pci_dev *pdev) otx2_pfaf_mbox_destroy(priv); pci_free_irq_vectors(priv->pdev); pci_set_drvdata(pdev, NULL); - pci_release_regions(pdev); } static struct pci_driver rvu_rep_driver = { |