diff options
Diffstat (limited to 'drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c')
| -rw-r--r-- | drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c | 2075 |
1 files changed, 1609 insertions, 466 deletions
diff --git a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c index 7ff147e89426..8d029addddad 100644 --- a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c +++ b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c @@ -1,6 +1,6 @@ // SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause) /* Copyright 2014-2016 Freescale Semiconductor Inc. - * Copyright 2016-2019 NXP + * Copyright 2016-2020 NXP */ #include <linux/init.h> #include <linux/module.h> @@ -11,11 +11,14 @@ #include <linux/msi.h> #include <linux/kthread.h> #include <linux/iommu.h> -#include <linux/net_tstamp.h> #include <linux/fsl/mc.h> #include <linux/bpf.h> #include <linux/bpf_trace.h> +#include <linux/fsl/ptp_qoriq.h> +#include <linux/ptp_classify.h> +#include <net/pkt_cls.h> #include <net/sock.h> +#include <net/tso.h> #include "dpaa2-eth.h" @@ -29,6 +32,78 @@ MODULE_LICENSE("Dual BSD/GPL"); MODULE_AUTHOR("Freescale Semiconductor, Inc"); MODULE_DESCRIPTION("Freescale DPAA2 Ethernet Driver"); +struct ptp_qoriq *dpaa2_ptp; +EXPORT_SYMBOL(dpaa2_ptp); + +static void dpaa2_eth_detect_features(struct dpaa2_eth_priv *priv) +{ + priv->features = 0; + + if (dpaa2_eth_cmp_dpni_ver(priv, DPNI_PTP_ONESTEP_VER_MAJOR, + DPNI_PTP_ONESTEP_VER_MINOR) >= 0) + priv->features |= DPAA2_ETH_FEATURE_ONESTEP_CFG_DIRECT; +} + +static void dpaa2_update_ptp_onestep_indirect(struct dpaa2_eth_priv *priv, + u32 offset, u8 udp) +{ + struct dpni_single_step_cfg cfg; + + cfg.en = 1; + cfg.ch_update = udp; + cfg.offset = offset; + cfg.peer_delay = 0; + + if (dpni_set_single_step_cfg(priv->mc_io, 0, priv->mc_token, &cfg)) + WARN_ONCE(1, "Failed to set single step register"); +} + +static void dpaa2_update_ptp_onestep_direct(struct dpaa2_eth_priv *priv, + u32 offset, u8 udp) +{ + u32 val = 0; + + val = DPAA2_PTP_SINGLE_STEP_ENABLE | + DPAA2_PTP_SINGLE_CORRECTION_OFF(offset); + + if (udp) + val |= DPAA2_PTP_SINGLE_STEP_CH; + + if (priv->onestep_reg_base) + writel(val, priv->onestep_reg_base); +} + +static void dpaa2_ptp_onestep_reg_update_method(struct dpaa2_eth_priv *priv) +{ + struct device *dev = priv->net_dev->dev.parent; + struct dpni_single_step_cfg ptp_cfg; + + priv->dpaa2_set_onestep_params_cb = dpaa2_update_ptp_onestep_indirect; + + if (!(priv->features & DPAA2_ETH_FEATURE_ONESTEP_CFG_DIRECT)) + return; + + if (dpni_get_single_step_cfg(priv->mc_io, 0, + priv->mc_token, &ptp_cfg)) { + dev_err(dev, "dpni_get_single_step_cfg cannot retrieve onestep reg, falling back to indirect update\n"); + return; + } + + if (!ptp_cfg.ptp_onestep_reg_base) { + dev_err(dev, "1588 onestep reg not available, falling back to indirect update\n"); + return; + } + + priv->onestep_reg_base = ioremap(ptp_cfg.ptp_onestep_reg_base, + sizeof(u32)); + if (!priv->onestep_reg_base) { + dev_err(dev, "1588 onestep reg cannot be mapped, falling back to indirect update\n"); + return; + } + + priv->dpaa2_set_onestep_params_cb = dpaa2_update_ptp_onestep_direct; +} + static void *dpaa2_iova_to_virt(struct iommu_domain *domain, dma_addr_t iova_addr) { @@ -39,9 +114,9 @@ static void *dpaa2_iova_to_virt(struct iommu_domain *domain, return phys_to_virt(phys_addr); } -static void validate_rx_csum(struct dpaa2_eth_priv *priv, - u32 fd_status, - struct sk_buff *skb) +static void dpaa2_eth_validate_rx_csum(struct dpaa2_eth_priv *priv, + u32 fd_status, + struct sk_buff *skb) { skb_checksum_none_assert(skb); @@ -61,9 +136,9 @@ static void validate_rx_csum(struct dpaa2_eth_priv *priv, /* Free a received FD. * Not to be used for Tx conf FDs or on any other paths. */ -static void free_rx_fd(struct dpaa2_eth_priv *priv, - const struct dpaa2_fd *fd, - void *vaddr) +static void dpaa2_eth_free_rx_fd(struct dpaa2_eth_priv *priv, + const struct dpaa2_fd *fd, + void *vaddr) { struct device *dev = priv->net_dev->dev.parent; dma_addr_t addr = dpaa2_fd_get_addr(fd); @@ -86,7 +161,7 @@ static void free_rx_fd(struct dpaa2_eth_priv *priv, for (i = 1; i < DPAA2_ETH_MAX_SG_ENTRIES; i++) { addr = dpaa2_sg_get_addr(&sgt[i]); sg_vaddr = dpaa2_iova_to_virt(priv->iommu_domain, addr); - dma_unmap_page(dev, addr, DPAA2_ETH_RX_BUF_SIZE, + dma_unmap_page(dev, addr, priv->rx_buf_size, DMA_BIDIRECTIONAL); free_pages((unsigned long)sg_vaddr, 0); @@ -99,9 +174,9 @@ free_buf: } /* Build a linear skb based on a single-buffer frame descriptor */ -static struct sk_buff *build_linear_skb(struct dpaa2_eth_channel *ch, - const struct dpaa2_fd *fd, - void *fd_vaddr) +static struct sk_buff *dpaa2_eth_build_linear_skb(struct dpaa2_eth_channel *ch, + const struct dpaa2_fd *fd, + void *fd_vaddr) { struct sk_buff *skb = NULL; u16 fd_offset = dpaa2_fd_get_offset(fd); @@ -120,9 +195,9 @@ static struct sk_buff *build_linear_skb(struct dpaa2_eth_channel *ch, } /* Build a non linear (fragmented) skb based on a S/G table */ -static struct sk_buff *build_frag_skb(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *ch, - struct dpaa2_sg_entry *sgt) +static struct sk_buff *dpaa2_eth_build_frag_skb(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + struct dpaa2_sg_entry *sgt) { struct sk_buff *skb = NULL; struct device *dev = priv->net_dev->dev.parent; @@ -144,7 +219,7 @@ static struct sk_buff *build_frag_skb(struct dpaa2_eth_priv *priv, /* Get the address and length from the S/G entry */ sg_addr = dpaa2_sg_get_addr(sge); sg_vaddr = dpaa2_iova_to_virt(priv->iommu_domain, sg_addr); - dma_unmap_page(dev, sg_addr, DPAA2_ETH_RX_BUF_SIZE, + dma_unmap_page(dev, sg_addr, priv->rx_buf_size, DMA_BIDIRECTIONAL); sg_length = dpaa2_sg_get_len(sge); @@ -185,7 +260,7 @@ static struct sk_buff *build_frag_skb(struct dpaa2_eth_priv *priv, (page_address(page) - page_address(head_page)); skb_add_rx_frag(skb, i - 1, head_page, page_offset, - sg_length, DPAA2_ETH_RX_BUF_SIZE); + sg_length, priv->rx_buf_size); } if (dpaa2_sg_is_final(sge)) @@ -203,7 +278,8 @@ static struct sk_buff *build_frag_skb(struct dpaa2_eth_priv *priv, /* Free buffers acquired from the buffer pool or which were meant to * be released in the pool */ -static void free_bufs(struct dpaa2_eth_priv *priv, u64 *buf_array, int count) +static void dpaa2_eth_free_bufs(struct dpaa2_eth_priv *priv, u64 *buf_array, + int count) { struct device *dev = priv->net_dev->dev.parent; void *vaddr; @@ -211,46 +287,105 @@ static void free_bufs(struct dpaa2_eth_priv *priv, u64 *buf_array, int count) for (i = 0; i < count; i++) { vaddr = dpaa2_iova_to_virt(priv->iommu_domain, buf_array[i]); - dma_unmap_page(dev, buf_array[i], DPAA2_ETH_RX_BUF_SIZE, + dma_unmap_page(dev, buf_array[i], priv->rx_buf_size, DMA_BIDIRECTIONAL); free_pages((unsigned long)vaddr, 0); } } -static void xdp_release_buf(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *ch, - dma_addr_t addr) +static void dpaa2_eth_recycle_buf(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + dma_addr_t addr) { int retries = 0; int err; - ch->xdp.drop_bufs[ch->xdp.drop_cnt++] = addr; - if (ch->xdp.drop_cnt < DPAA2_ETH_BUFS_PER_CMD) + ch->recycled_bufs[ch->recycled_bufs_cnt++] = addr; + if (ch->recycled_bufs_cnt < DPAA2_ETH_BUFS_PER_CMD) return; while ((err = dpaa2_io_service_release(ch->dpio, priv->bpid, - ch->xdp.drop_bufs, - ch->xdp.drop_cnt)) == -EBUSY) { + ch->recycled_bufs, + ch->recycled_bufs_cnt)) == -EBUSY) { if (retries++ >= DPAA2_ETH_SWP_BUSY_RETRIES) break; cpu_relax(); } if (err) { - free_bufs(priv, ch->xdp.drop_bufs, ch->xdp.drop_cnt); - ch->buf_count -= ch->xdp.drop_cnt; + dpaa2_eth_free_bufs(priv, ch->recycled_bufs, ch->recycled_bufs_cnt); + ch->buf_count -= ch->recycled_bufs_cnt; } - ch->xdp.drop_cnt = 0; + ch->recycled_bufs_cnt = 0; } -static int xdp_enqueue(struct dpaa2_eth_priv *priv, struct dpaa2_fd *fd, - void *buf_start, u16 queue_id) +static int dpaa2_eth_xdp_flush(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_fq *fq, + struct dpaa2_eth_xdp_fds *xdp_fds) +{ + int total_enqueued = 0, retries = 0, enqueued; + struct dpaa2_eth_drv_stats *percpu_extras; + int num_fds, err, max_retries; + struct dpaa2_fd *fds; + + percpu_extras = this_cpu_ptr(priv->percpu_extras); + + /* try to enqueue all the FDs until the max number of retries is hit */ + fds = xdp_fds->fds; + num_fds = xdp_fds->num; + max_retries = num_fds * DPAA2_ETH_ENQUEUE_RETRIES; + while (total_enqueued < num_fds && retries < max_retries) { + err = priv->enqueue(priv, fq, &fds[total_enqueued], + 0, num_fds - total_enqueued, &enqueued); + if (err == -EBUSY) { + percpu_extras->tx_portal_busy += ++retries; + continue; + } + total_enqueued += enqueued; + } + xdp_fds->num = 0; + + return total_enqueued; +} + +static void dpaa2_eth_xdp_tx_flush(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + struct dpaa2_eth_fq *fq) +{ + struct rtnl_link_stats64 *percpu_stats; + struct dpaa2_fd *fds; + int enqueued, i; + + percpu_stats = this_cpu_ptr(priv->percpu_stats); + + // enqueue the array of XDP_TX frames + enqueued = dpaa2_eth_xdp_flush(priv, fq, &fq->xdp_tx_fds); + + /* update statistics */ + percpu_stats->tx_packets += enqueued; + fds = fq->xdp_tx_fds.fds; + for (i = 0; i < enqueued; i++) { + percpu_stats->tx_bytes += dpaa2_fd_get_len(&fds[i]); + ch->stats.xdp_tx++; + } + for (i = enqueued; i < fq->xdp_tx_fds.num; i++) { + dpaa2_eth_recycle_buf(priv, ch, dpaa2_fd_get_addr(&fds[i])); + percpu_stats->tx_errors++; + ch->stats.xdp_tx_err++; + } + fq->xdp_tx_fds.num = 0; +} + +static void dpaa2_eth_xdp_enqueue(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + struct dpaa2_fd *fd, + void *buf_start, u16 queue_id) { - struct dpaa2_eth_fq *fq; struct dpaa2_faead *faead; + struct dpaa2_fd *dest_fd; + struct dpaa2_eth_fq *fq; u32 ctrl, frc; - int i, err; /* Mark the egress frame hardware annotation area as valid */ frc = dpaa2_fd_get_frc(fd); @@ -267,40 +402,34 @@ static int xdp_enqueue(struct dpaa2_eth_priv *priv, struct dpaa2_fd *fd, faead->conf_fqid = 0; fq = &priv->fq[queue_id]; - for (i = 0; i < DPAA2_ETH_ENQUEUE_RETRIES; i++) { - err = priv->enqueue(priv, fq, fd, 0); - if (err != -EBUSY) - break; - } + dest_fd = &fq->xdp_tx_fds.fds[fq->xdp_tx_fds.num++]; + memcpy(dest_fd, fd, sizeof(*dest_fd)); - return err; + if (fq->xdp_tx_fds.num < DEV_MAP_BULK_SIZE) + return; + + dpaa2_eth_xdp_tx_flush(priv, ch, fq); } -static u32 run_xdp(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *ch, - struct dpaa2_eth_fq *rx_fq, - struct dpaa2_fd *fd, void *vaddr) +static u32 dpaa2_eth_run_xdp(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + struct dpaa2_eth_fq *rx_fq, + struct dpaa2_fd *fd, void *vaddr) { dma_addr_t addr = dpaa2_fd_get_addr(fd); - struct rtnl_link_stats64 *percpu_stats; struct bpf_prog *xdp_prog; struct xdp_buff xdp; u32 xdp_act = XDP_PASS; - int err; - - percpu_stats = this_cpu_ptr(priv->percpu_stats); - - rcu_read_lock(); + int err, offset; xdp_prog = READ_ONCE(ch->xdp.prog); if (!xdp_prog) goto out; - xdp.data = vaddr + dpaa2_fd_get_offset(fd); - xdp.data_end = xdp.data + dpaa2_fd_get_len(fd); - xdp.data_hard_start = xdp.data - XDP_PACKET_HEADROOM; - xdp_set_data_meta_invalid(&xdp); - xdp.rxq = &ch->xdp_rxq; + offset = dpaa2_fd_get_offset(fd) - XDP_PACKET_HEADROOM; + xdp_init_buff(&xdp, DPAA2_ETH_RX_BUF_RAW_SIZE - offset, &ch->xdp_rxq); + xdp_prepare_buff(&xdp, vaddr + offset, XDP_PACKET_HEADROOM, + dpaa2_fd_get_len(fd), false); xdp_act = bpf_prog_run_xdp(xdp_prog, &xdp); @@ -312,46 +441,79 @@ static u32 run_xdp(struct dpaa2_eth_priv *priv, case XDP_PASS: break; case XDP_TX: - err = xdp_enqueue(priv, fd, vaddr, rx_fq->flowid); - if (err) { - xdp_release_buf(priv, ch, addr); - percpu_stats->tx_errors++; - ch->stats.xdp_tx_err++; - } else { - percpu_stats->tx_packets++; - percpu_stats->tx_bytes += dpaa2_fd_get_len(fd); - ch->stats.xdp_tx++; - } + dpaa2_eth_xdp_enqueue(priv, ch, fd, vaddr, rx_fq->flowid); break; default: - bpf_warn_invalid_xdp_action(xdp_act); - /* fall through */ + bpf_warn_invalid_xdp_action(priv->net_dev, xdp_prog, xdp_act); + fallthrough; case XDP_ABORTED: trace_xdp_exception(priv->net_dev, xdp_prog, xdp_act); - /* fall through */ + fallthrough; case XDP_DROP: - xdp_release_buf(priv, ch, addr); + dpaa2_eth_recycle_buf(priv, ch, addr); ch->stats.xdp_drop++; break; case XDP_REDIRECT: dma_unmap_page(priv->net_dev->dev.parent, addr, - DPAA2_ETH_RX_BUF_SIZE, DMA_BIDIRECTIONAL); + priv->rx_buf_size, DMA_BIDIRECTIONAL); ch->buf_count--; + + /* Allow redirect use of full headroom */ xdp.data_hard_start = vaddr; + xdp.frame_sz = DPAA2_ETH_RX_BUF_RAW_SIZE; + err = xdp_do_redirect(priv->net_dev, &xdp, xdp_prog); - if (unlikely(err)) + if (unlikely(err)) { + addr = dma_map_page(priv->net_dev->dev.parent, + virt_to_page(vaddr), 0, + priv->rx_buf_size, DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(priv->net_dev->dev.parent, addr))) { + free_pages((unsigned long)vaddr, 0); + } else { + ch->buf_count++; + dpaa2_eth_recycle_buf(priv, ch, addr); + } ch->stats.xdp_drop++; - else + } else { ch->stats.xdp_redirect++; + } break; } ch->xdp.res |= xdp_act; out: - rcu_read_unlock(); return xdp_act; } +static struct sk_buff *dpaa2_eth_copybreak(struct dpaa2_eth_channel *ch, + const struct dpaa2_fd *fd, + void *fd_vaddr) +{ + u16 fd_offset = dpaa2_fd_get_offset(fd); + struct dpaa2_eth_priv *priv = ch->priv; + u32 fd_length = dpaa2_fd_get_len(fd); + struct sk_buff *skb = NULL; + unsigned int skb_len; + + if (fd_length > priv->rx_copybreak) + return NULL; + + skb_len = fd_length + dpaa2_eth_needed_headroom(NULL); + + skb = napi_alloc_skb(&ch->napi, skb_len); + if (!skb) + return NULL; + + skb_reserve(skb, dpaa2_eth_needed_headroom(NULL)); + skb_put(skb, fd_length); + + memcpy(skb->data, fd_vaddr + fd_offset, fd_length); + + dpaa2_eth_recycle_buf(priv, ch, dpaa2_fd_get_addr(fd)); + + return skb; +} + /* Main Rx frame processing routine */ static void dpaa2_eth_rx(struct dpaa2_eth_priv *priv, struct dpaa2_eth_channel *ch, @@ -374,7 +536,7 @@ static void dpaa2_eth_rx(struct dpaa2_eth_priv *priv, trace_dpaa2_rx_fd(priv->net_dev, fd); vaddr = dpaa2_iova_to_virt(priv->iommu_domain, addr); - dma_sync_single_for_cpu(dev, addr, DPAA2_ETH_RX_BUF_SIZE, + dma_sync_single_for_cpu(dev, addr, priv->rx_buf_size, DMA_BIDIRECTIONAL); fas = dpaa2_get_fas(vaddr, false); @@ -386,22 +548,25 @@ static void dpaa2_eth_rx(struct dpaa2_eth_priv *priv, percpu_extras = this_cpu_ptr(priv->percpu_extras); if (fd_format == dpaa2_fd_single) { - xdp_act = run_xdp(priv, ch, fq, (struct dpaa2_fd *)fd, vaddr); + xdp_act = dpaa2_eth_run_xdp(priv, ch, fq, (struct dpaa2_fd *)fd, vaddr); if (xdp_act != XDP_PASS) { percpu_stats->rx_packets++; percpu_stats->rx_bytes += dpaa2_fd_get_len(fd); return; } - dma_unmap_page(dev, addr, DPAA2_ETH_RX_BUF_SIZE, - DMA_BIDIRECTIONAL); - skb = build_linear_skb(ch, fd, vaddr); + skb = dpaa2_eth_copybreak(ch, fd, vaddr); + if (!skb) { + dma_unmap_page(dev, addr, priv->rx_buf_size, + DMA_BIDIRECTIONAL); + skb = dpaa2_eth_build_linear_skb(ch, fd, vaddr); + } } else if (fd_format == dpaa2_fd_sg) { WARN_ON(priv->xdp_prog); - dma_unmap_page(dev, addr, DPAA2_ETH_RX_BUF_SIZE, + dma_unmap_page(dev, addr, priv->rx_buf_size, DMA_BIDIRECTIONAL); - skb = build_frag_skb(priv, ch, buf_data); + skb = dpaa2_eth_build_frag_skb(priv, ch, buf_data); free_pages((unsigned long)vaddr, 0); percpu_extras->rx_sg_frames++; percpu_extras->rx_sg_bytes += dpaa2_fd_get_len(fd); @@ -430,7 +595,7 @@ static void dpaa2_eth_rx(struct dpaa2_eth_priv *priv, /* Check if we need to validate the L4 csum */ if (likely(dpaa2_fd_get_frc(fd) & DPAA2_FD_FRC_FASV)) { status = le32_to_cpu(fas->status); - validate_rx_csum(priv, status, skb); + dpaa2_eth_validate_rx_csum(priv, status, skb); } skb->protocol = eth_type_trans(skb, priv->net_dev); @@ -438,25 +603,78 @@ static void dpaa2_eth_rx(struct dpaa2_eth_priv *priv, percpu_stats->rx_packets++; percpu_stats->rx_bytes += dpaa2_fd_get_len(fd); + ch->stats.bytes_per_cdan += dpaa2_fd_get_len(fd); list_add_tail(&skb->list, ch->rx_list); return; err_build_skb: - free_rx_fd(priv, fd, vaddr); + dpaa2_eth_free_rx_fd(priv, fd, vaddr); err_frame_format: percpu_stats->rx_dropped++; } +/* Processing of Rx frames received on the error FQ + * We check and print the error bits and then free the frame + */ +static void dpaa2_eth_rx_err(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + const struct dpaa2_fd *fd, + struct dpaa2_eth_fq *fq __always_unused) +{ + struct device *dev = priv->net_dev->dev.parent; + dma_addr_t addr = dpaa2_fd_get_addr(fd); + u8 fd_format = dpaa2_fd_get_format(fd); + struct rtnl_link_stats64 *percpu_stats; + struct dpaa2_eth_trap_item *trap_item; + struct dpaa2_fapr *fapr; + struct sk_buff *skb; + void *buf_data; + void *vaddr; + + vaddr = dpaa2_iova_to_virt(priv->iommu_domain, addr); + dma_sync_single_for_cpu(dev, addr, priv->rx_buf_size, + DMA_BIDIRECTIONAL); + + buf_data = vaddr + dpaa2_fd_get_offset(fd); + + if (fd_format == dpaa2_fd_single) { + dma_unmap_page(dev, addr, priv->rx_buf_size, + DMA_BIDIRECTIONAL); + skb = dpaa2_eth_build_linear_skb(ch, fd, vaddr); + } else if (fd_format == dpaa2_fd_sg) { + dma_unmap_page(dev, addr, priv->rx_buf_size, + DMA_BIDIRECTIONAL); + skb = dpaa2_eth_build_frag_skb(priv, ch, buf_data); + free_pages((unsigned long)vaddr, 0); + } else { + /* We don't support any other format */ + dpaa2_eth_free_rx_fd(priv, fd, vaddr); + goto err_frame_format; + } + + fapr = dpaa2_get_fapr(vaddr, false); + trap_item = dpaa2_eth_dl_get_trap(priv, fapr); + if (trap_item) + devlink_trap_report(priv->devlink, skb, trap_item->trap_ctx, + &priv->devlink_port, NULL); + consume_skb(skb); + +err_frame_format: + percpu_stats = this_cpu_ptr(priv->percpu_stats); + percpu_stats->rx_errors++; + ch->buf_count--; +} + /* Consume all frames pull-dequeued into the store. This is the simplest way to * make sure we don't accidentally issue another volatile dequeue which would * overwrite (leak) frames already in the store. * * Observance of NAPI budget is not our concern, leaving that to the caller. */ -static int consume_frames(struct dpaa2_eth_channel *ch, - struct dpaa2_eth_fq **src) +static int dpaa2_eth_consume_frames(struct dpaa2_eth_channel *ch, + struct dpaa2_eth_fq **src) { struct dpaa2_eth_priv *priv = ch->priv; struct dpaa2_eth_fq *fq = NULL; @@ -493,6 +711,8 @@ static int consume_frames(struct dpaa2_eth_channel *ch, return 0; fq->stats.frames += cleaned; + ch->stats.frames += cleaned; + ch->stats.frames_per_cdan += cleaned; /* A dequeue operation only pulls frames from a single queue * into the store. Return the frame queue as an out param. @@ -503,11 +723,56 @@ static int consume_frames(struct dpaa2_eth_channel *ch, return cleaned; } +static int dpaa2_eth_ptp_parse(struct sk_buff *skb, + u8 *msgtype, u8 *twostep, u8 *udp, + u16 *correction_offset, + u16 *origintimestamp_offset) +{ + unsigned int ptp_class; + struct ptp_header *hdr; + unsigned int type; + u8 *base; + + ptp_class = ptp_classify_raw(skb); + if (ptp_class == PTP_CLASS_NONE) + return -EINVAL; + + hdr = ptp_parse_header(skb, ptp_class); + if (!hdr) + return -EINVAL; + + *msgtype = ptp_get_msgtype(hdr, ptp_class); + *twostep = hdr->flag_field[0] & 0x2; + + type = ptp_class & PTP_CLASS_PMASK; + if (type == PTP_CLASS_IPV4 || + type == PTP_CLASS_IPV6) + *udp = 1; + else + *udp = 0; + + base = skb_mac_header(skb); + *correction_offset = (u8 *)&hdr->correction - base; + *origintimestamp_offset = (u8 *)hdr + sizeof(struct ptp_header) - base; + + return 0; +} + /* Configure the egress frame annotation for timestamp update */ -static void enable_tx_tstamp(struct dpaa2_fd *fd, void *buf_start) +static void dpaa2_eth_enable_tx_tstamp(struct dpaa2_eth_priv *priv, + struct dpaa2_fd *fd, + void *buf_start, + struct sk_buff *skb) { + struct ptp_tstamp origin_timestamp; + u8 msgtype, twostep, udp; struct dpaa2_faead *faead; + struct dpaa2_fas *fas; + struct timespec64 ts; + u16 offset1, offset2; u32 ctrl, frc; + __le64 *ns; + u8 *data; /* Mark the egress frame annotation area as valid */ frc = dpaa2_fd_get_frc(fd); @@ -523,12 +788,83 @@ static void enable_tx_tstamp(struct dpaa2_fd *fd, void *buf_start) ctrl = DPAA2_FAEAD_A2V | DPAA2_FAEAD_UPDV | DPAA2_FAEAD_UPD; faead = dpaa2_get_faead(buf_start, true); faead->ctrl = cpu_to_le32(ctrl); + + if (skb->cb[0] == TX_TSTAMP_ONESTEP_SYNC) { + if (dpaa2_eth_ptp_parse(skb, &msgtype, &twostep, &udp, + &offset1, &offset2) || + msgtype != PTP_MSGTYPE_SYNC || twostep) { + WARN_ONCE(1, "Bad packet for one-step timestamping\n"); + return; + } + + /* Mark the frame annotation status as valid */ + frc = dpaa2_fd_get_frc(fd); + dpaa2_fd_set_frc(fd, frc | DPAA2_FD_FRC_FASV); + + /* Mark the PTP flag for one step timestamping */ + fas = dpaa2_get_fas(buf_start, true); + fas->status = cpu_to_le32(DPAA2_FAS_PTP); + + dpaa2_ptp->caps.gettime64(&dpaa2_ptp->caps, &ts); + ns = dpaa2_get_ts(buf_start, true); + *ns = cpu_to_le64(timespec64_to_ns(&ts) / + DPAA2_PTP_CLK_PERIOD_NS); + + /* Update current time to PTP message originTimestamp field */ + ns_to_ptp_tstamp(&origin_timestamp, le64_to_cpup(ns)); + data = skb_mac_header(skb); + *(__be16 *)(data + offset2) = htons(origin_timestamp.sec_msb); + *(__be32 *)(data + offset2 + 2) = + htonl(origin_timestamp.sec_lsb); + *(__be32 *)(data + offset2 + 6) = htonl(origin_timestamp.nsec); + + if (priv->ptp_correction_off == offset1) + return; + + priv->dpaa2_set_onestep_params_cb(priv, offset1, udp); + priv->ptp_correction_off = offset1; + + } +} + +static void *dpaa2_eth_sgt_get(struct dpaa2_eth_priv *priv) +{ + struct dpaa2_eth_sgt_cache *sgt_cache; + void *sgt_buf = NULL; + int sgt_buf_size; + + sgt_cache = this_cpu_ptr(priv->sgt_cache); + sgt_buf_size = priv->tx_data_offset + + DPAA2_ETH_SG_ENTRIES_MAX * sizeof(struct dpaa2_sg_entry); + + if (sgt_cache->count == 0) + sgt_buf = napi_alloc_frag_align(sgt_buf_size, DPAA2_ETH_TX_BUF_ALIGN); + else + sgt_buf = sgt_cache->buf[--sgt_cache->count]; + if (!sgt_buf) + return NULL; + + memset(sgt_buf, 0, sgt_buf_size); + + return sgt_buf; +} + +static void dpaa2_eth_sgt_recycle(struct dpaa2_eth_priv *priv, void *sgt_buf) +{ + struct dpaa2_eth_sgt_cache *sgt_cache; + + sgt_cache = this_cpu_ptr(priv->sgt_cache); + if (sgt_cache->count >= DPAA2_ETH_SGT_CACHE_SIZE) + skb_free_frag(sgt_buf); + else + sgt_cache->buf[sgt_cache->count++] = sgt_buf; } /* Create a frame descriptor based on a fragmented skb */ -static int build_sg_fd(struct dpaa2_eth_priv *priv, - struct sk_buff *skb, - struct dpaa2_fd *fd) +static int dpaa2_eth_build_sg_fd(struct dpaa2_eth_priv *priv, + struct sk_buff *skb, + struct dpaa2_fd *fd, + void **swa_addr) { struct device *dev = priv->net_dev->dev.parent; void *sgt_buf = NULL; @@ -550,12 +886,16 @@ static int build_sg_fd(struct dpaa2_eth_priv *priv, if (unlikely(PAGE_SIZE / sizeof(struct scatterlist) < nr_frags + 1)) return -EINVAL; - scl = kcalloc(nr_frags + 1, sizeof(struct scatterlist), GFP_ATOMIC); + scl = kmalloc_array(nr_frags + 1, sizeof(struct scatterlist), GFP_ATOMIC); if (unlikely(!scl)) return -ENOMEM; sg_init_table(scl, nr_frags + 1); num_sg = skb_to_sgvec(skb, scl, 0, skb->len); + if (unlikely(num_sg < 0)) { + err = -ENOMEM; + goto dma_map_sg_failed; + } num_dma_bufs = dma_map_sg(dev, scl, num_sg, DMA_BIDIRECTIONAL); if (unlikely(!num_dma_bufs)) { err = -ENOMEM; @@ -565,13 +905,11 @@ static int build_sg_fd(struct dpaa2_eth_priv *priv, /* Prepare the HW SGT structure */ sgt_buf_size = priv->tx_data_offset + sizeof(struct dpaa2_sg_entry) * num_dma_bufs; - sgt_buf = napi_alloc_frag(sgt_buf_size + DPAA2_ETH_TX_BUF_ALIGN); + sgt_buf = dpaa2_eth_sgt_get(priv); if (unlikely(!sgt_buf)) { err = -ENOMEM; goto sgt_buf_alloc_failed; } - sgt_buf = PTR_ALIGN(sgt_buf, DPAA2_ETH_TX_BUF_ALIGN); - memset(sgt_buf, 0, sgt_buf_size); sgt = (struct dpaa2_sg_entry *)(sgt_buf + priv->tx_data_offset); @@ -593,6 +931,7 @@ static int build_sg_fd(struct dpaa2_eth_priv *priv, * skb backpointer in the software annotation area. We'll need * all of them on Tx Conf. */ + *swa_addr = (void *)sgt_buf; swa = (struct dpaa2_eth_swa *)sgt_buf; swa->type = DPAA2_ETH_SWA_SG; swa->sg.skb = skb; @@ -606,19 +945,17 @@ static int build_sg_fd(struct dpaa2_eth_priv *priv, err = -ENOMEM; goto dma_map_single_failed; } + memset(fd, 0, sizeof(struct dpaa2_fd)); dpaa2_fd_set_offset(fd, priv->tx_data_offset); dpaa2_fd_set_format(fd, dpaa2_fd_sg); dpaa2_fd_set_addr(fd, addr); dpaa2_fd_set_len(fd, skb->len); dpaa2_fd_set_ctrl(fd, FD_CTRL_PTA); - if (priv->tx_tstamp && skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP) - enable_tx_tstamp(fd, sgt_buf); - return 0; dma_map_single_failed: - skb_free_frag(sgt_buf); + dpaa2_eth_sgt_recycle(priv, sgt_buf); sgt_buf_alloc_failed: dma_unmap_sg(dev, scl, num_sg, DMA_BIDIRECTIONAL); dma_map_sg_failed: @@ -626,17 +963,86 @@ dma_map_sg_failed: return err; } +/* Create a SG frame descriptor based on a linear skb. + * + * This function is used on the Tx path when the skb headroom is not large + * enough for the HW requirements, thus instead of realloc-ing the skb we + * create a SG frame descriptor with only one entry. + */ +static int dpaa2_eth_build_sg_fd_single_buf(struct dpaa2_eth_priv *priv, + struct sk_buff *skb, + struct dpaa2_fd *fd, + void **swa_addr) +{ + struct device *dev = priv->net_dev->dev.parent; + struct dpaa2_sg_entry *sgt; + struct dpaa2_eth_swa *swa; + dma_addr_t addr, sgt_addr; + void *sgt_buf = NULL; + int sgt_buf_size; + int err; + + /* Prepare the HW SGT structure */ + sgt_buf_size = priv->tx_data_offset + sizeof(struct dpaa2_sg_entry); + sgt_buf = dpaa2_eth_sgt_get(priv); + if (unlikely(!sgt_buf)) + return -ENOMEM; + sgt = (struct dpaa2_sg_entry *)(sgt_buf + priv->tx_data_offset); + + addr = dma_map_single(dev, skb->data, skb->len, DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(dev, addr))) { + err = -ENOMEM; + goto data_map_failed; + } + + /* Fill in the HW SGT structure */ + dpaa2_sg_set_addr(sgt, addr); + dpaa2_sg_set_len(sgt, skb->len); + dpaa2_sg_set_final(sgt, true); + + /* Store the skb backpointer in the SGT buffer */ + *swa_addr = (void *)sgt_buf; + swa = (struct dpaa2_eth_swa *)sgt_buf; + swa->type = DPAA2_ETH_SWA_SINGLE; + swa->single.skb = skb; + swa->single.sgt_size = sgt_buf_size; + + /* Separately map the SGT buffer */ + sgt_addr = dma_map_single(dev, sgt_buf, sgt_buf_size, DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(dev, sgt_addr))) { + err = -ENOMEM; + goto sgt_map_failed; + } + + memset(fd, 0, sizeof(struct dpaa2_fd)); + dpaa2_fd_set_offset(fd, priv->tx_data_offset); + dpaa2_fd_set_format(fd, dpaa2_fd_sg); + dpaa2_fd_set_addr(fd, sgt_addr); + dpaa2_fd_set_len(fd, skb->len); + dpaa2_fd_set_ctrl(fd, FD_CTRL_PTA); + + return 0; + +sgt_map_failed: + dma_unmap_single(dev, addr, skb->len, DMA_BIDIRECTIONAL); +data_map_failed: + dpaa2_eth_sgt_recycle(priv, sgt_buf); + + return err; +} + /* Create a frame descriptor based on a linear skb */ -static int build_single_fd(struct dpaa2_eth_priv *priv, - struct sk_buff *skb, - struct dpaa2_fd *fd) +static int dpaa2_eth_build_single_fd(struct dpaa2_eth_priv *priv, + struct sk_buff *skb, + struct dpaa2_fd *fd, + void **swa_addr) { struct device *dev = priv->net_dev->dev.parent; u8 *buffer_start, *aligned_start; struct dpaa2_eth_swa *swa; dma_addr_t addr; - buffer_start = skb->data - dpaa2_eth_needed_headroom(priv, skb); + buffer_start = skb->data - dpaa2_eth_needed_headroom(skb); /* If there's enough room to align the FD address, do it. * It will help hardware optimize accesses. @@ -650,6 +1056,7 @@ static int build_single_fd(struct dpaa2_eth_priv *priv, * (in the private data area) such that we can release it * on Tx confirm */ + *swa_addr = (void *)buffer_start; swa = (struct dpaa2_eth_swa *)buffer_start; swa->type = DPAA2_ETH_SWA_SINGLE; swa->single.skb = skb; @@ -660,15 +1067,13 @@ static int build_single_fd(struct dpaa2_eth_priv *priv, if (unlikely(dma_mapping_error(dev, addr))) return -ENOMEM; + memset(fd, 0, sizeof(struct dpaa2_fd)); dpaa2_fd_set_addr(fd, addr); dpaa2_fd_set_offset(fd, (u16)(skb->data - buffer_start)); dpaa2_fd_set_len(fd, skb->len); dpaa2_fd_set_format(fd, dpaa2_fd_single); dpaa2_fd_set_ctrl(fd, FD_CTRL_PTA); - if (priv->tx_tstamp && skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP) - enable_tx_tstamp(fd, buffer_start); - return 0; } @@ -679,17 +1084,21 @@ static int build_single_fd(struct dpaa2_eth_priv *priv, * This can be called either from dpaa2_eth_tx_conf() or on the error path of * dpaa2_eth_tx(). */ -static void free_tx_fd(const struct dpaa2_eth_priv *priv, - struct dpaa2_eth_fq *fq, - const struct dpaa2_fd *fd, bool in_napi) +static void dpaa2_eth_free_tx_fd(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_fq *fq, + const struct dpaa2_fd *fd, bool in_napi) { struct device *dev = priv->net_dev->dev.parent; - dma_addr_t fd_addr; + dma_addr_t fd_addr, sg_addr; struct sk_buff *skb = NULL; unsigned char *buffer_start; struct dpaa2_eth_swa *swa; u8 fd_format = dpaa2_fd_get_format(fd); u32 fd_len = dpaa2_fd_get_len(fd); + struct dpaa2_sg_entry *sgt; + int should_free_skb = 1; + void *tso_hdr; + int i; fd_addr = dpaa2_fd_get_addr(fd); buffer_start = dpaa2_iova_to_virt(priv->iommu_domain, fd_addr); @@ -710,16 +1119,52 @@ static void free_tx_fd(const struct dpaa2_eth_priv *priv, DMA_BIDIRECTIONAL); } } else if (fd_format == dpaa2_fd_sg) { - skb = swa->sg.skb; + if (swa->type == DPAA2_ETH_SWA_SG) { + skb = swa->sg.skb; + + /* Unmap the scatterlist */ + dma_unmap_sg(dev, swa->sg.scl, swa->sg.num_sg, + DMA_BIDIRECTIONAL); + kfree(swa->sg.scl); + + /* Unmap the SGT buffer */ + dma_unmap_single(dev, fd_addr, swa->sg.sgt_size, + DMA_BIDIRECTIONAL); + } else if (swa->type == DPAA2_ETH_SWA_SW_TSO) { + skb = swa->tso.skb; + + sgt = (struct dpaa2_sg_entry *)(buffer_start + + priv->tx_data_offset); + + /* Unmap the SGT buffer */ + dma_unmap_single(dev, fd_addr, swa->tso.sgt_size, + DMA_BIDIRECTIONAL); - /* Unmap the scatterlist */ - dma_unmap_sg(dev, swa->sg.scl, swa->sg.num_sg, - DMA_BIDIRECTIONAL); - kfree(swa->sg.scl); + /* Unmap and free the header */ + tso_hdr = dpaa2_iova_to_virt(priv->iommu_domain, dpaa2_sg_get_addr(sgt)); + dma_unmap_single(dev, dpaa2_sg_get_addr(sgt), TSO_HEADER_SIZE, + DMA_TO_DEVICE); + kfree(tso_hdr); - /* Unmap the SGT buffer */ - dma_unmap_single(dev, fd_addr, swa->sg.sgt_size, - DMA_BIDIRECTIONAL); + /* Unmap the other SG entries for the data */ + for (i = 1; i < swa->tso.num_sg; i++) + dma_unmap_single(dev, dpaa2_sg_get_addr(&sgt[i]), + dpaa2_sg_get_len(&sgt[i]), DMA_TO_DEVICE); + + if (!swa->tso.is_last_fd) + should_free_skb = 0; + } else { + skb = swa->single.skb; + + /* Unmap the SGT Buffer */ + dma_unmap_single(dev, fd_addr, swa->single.sgt_size, + DMA_BIDIRECTIONAL); + + sgt = (struct dpaa2_sg_entry *)(buffer_start + + priv->tx_data_offset); + sg_addr = dpaa2_sg_get_addr(sgt); + dma_unmap_single(dev, sg_addr, skb->len, DMA_BIDIRECTIONAL); + } } else { netdev_dbg(priv->net_dev, "Invalid FD format\n"); return; @@ -736,60 +1181,197 @@ static void free_tx_fd(const struct dpaa2_eth_priv *priv, } /* Get the timestamp value */ - if (priv->tx_tstamp && skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP) { - struct skb_shared_hwtstamps shhwtstamps; - __le64 *ts = dpaa2_get_ts(buffer_start, true); - u64 ns; - - memset(&shhwtstamps, 0, sizeof(shhwtstamps)); - - ns = DPAA2_PTP_CLK_PERIOD_NS * le64_to_cpup(ts); - shhwtstamps.hwtstamp = ns_to_ktime(ns); - skb_tstamp_tx(skb, &shhwtstamps); + if (swa->type != DPAA2_ETH_SWA_SW_TSO) { + if (skb->cb[0] == TX_TSTAMP) { + struct skb_shared_hwtstamps shhwtstamps; + __le64 *ts = dpaa2_get_ts(buffer_start, true); + u64 ns; + + memset(&shhwtstamps, 0, sizeof(shhwtstamps)); + + ns = DPAA2_PTP_CLK_PERIOD_NS * le64_to_cpup(ts); + shhwtstamps.hwtstamp = ns_to_ktime(ns); + skb_tstamp_tx(skb, &shhwtstamps); + } else if (skb->cb[0] == TX_TSTAMP_ONESTEP_SYNC) { + mutex_unlock(&priv->onestep_tstamp_lock); + } } /* Free SGT buffer allocated on tx */ if (fd_format != dpaa2_fd_single) - skb_free_frag(buffer_start); + dpaa2_eth_sgt_recycle(priv, buffer_start); - /* Move on with skb release */ - napi_consume_skb(skb, in_napi); + /* Move on with skb release. If we are just confirming multiple FDs + * from the same TSO skb then only the last one will need to free the + * skb. + */ + if (should_free_skb) + napi_consume_skb(skb, in_napi); } -static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev) +static int dpaa2_eth_build_gso_fd(struct dpaa2_eth_priv *priv, + struct sk_buff *skb, struct dpaa2_fd *fd, + int *num_fds, u32 *total_fds_len) +{ + struct device *dev = priv->net_dev->dev.parent; + int hdr_len, total_len, data_left, fd_len; + int num_sge, err, i, sgt_buf_size; + struct dpaa2_fd *fd_start = fd; + struct dpaa2_sg_entry *sgt; + struct dpaa2_eth_swa *swa; + dma_addr_t sgt_addr, addr; + dma_addr_t tso_hdr_dma; + unsigned int index = 0; + struct tso_t tso; + char *tso_hdr; + void *sgt_buf; + + /* Initialize the TSO handler, and prepare the first payload */ + hdr_len = tso_start(skb, &tso); + *total_fds_len = 0; + + total_len = skb->len - hdr_len; + while (total_len > 0) { + /* Prepare the HW SGT structure for this frame */ + sgt_buf = dpaa2_eth_sgt_get(priv); + if (unlikely(!sgt_buf)) { + netdev_err(priv->net_dev, "dpaa2_eth_sgt_get() failed\n"); + err = -ENOMEM; + goto err_sgt_get; + } + sgt = (struct dpaa2_sg_entry *)(sgt_buf + priv->tx_data_offset); + + /* Determine the data length of this frame */ + data_left = min_t(int, skb_shinfo(skb)->gso_size, total_len); + total_len -= data_left; + fd_len = data_left + hdr_len; + + /* Prepare packet headers: MAC + IP + TCP */ + tso_hdr = kmalloc(TSO_HEADER_SIZE, GFP_ATOMIC); + if (!tso_hdr) { + err = -ENOMEM; + goto err_alloc_tso_hdr; + } + + tso_build_hdr(skb, tso_hdr, &tso, data_left, total_len == 0); + tso_hdr_dma = dma_map_single(dev, tso_hdr, TSO_HEADER_SIZE, DMA_TO_DEVICE); + if (dma_mapping_error(dev, tso_hdr_dma)) { + netdev_err(priv->net_dev, "dma_map_single(tso_hdr) failed\n"); + err = -ENOMEM; + goto err_map_tso_hdr; + } + + /* Setup the SG entry for the header */ + dpaa2_sg_set_addr(sgt, tso_hdr_dma); + dpaa2_sg_set_len(sgt, hdr_len); + dpaa2_sg_set_final(sgt, data_left <= 0); + + /* Compose the SG entries for each fragment of data */ + num_sge = 1; + while (data_left > 0) { + int size; + + /* Move to the next SG entry */ + sgt++; + size = min_t(int, tso.size, data_left); + + addr = dma_map_single(dev, tso.data, size, DMA_TO_DEVICE); + if (dma_mapping_error(dev, addr)) { + netdev_err(priv->net_dev, "dma_map_single(tso.data) failed\n"); + err = -ENOMEM; + goto err_map_data; + } + dpaa2_sg_set_addr(sgt, addr); + dpaa2_sg_set_len(sgt, size); + dpaa2_sg_set_final(sgt, size == data_left); + + num_sge++; + + /* Build the data for the __next__ fragment */ + data_left -= size; + tso_build_data(skb, &tso, size); + } + + /* Store the skb backpointer in the SGT buffer */ + sgt_buf_size = priv->tx_data_offset + num_sge * sizeof(struct dpaa2_sg_entry); + swa = (struct dpaa2_eth_swa *)sgt_buf; + swa->type = DPAA2_ETH_SWA_SW_TSO; + swa->tso.skb = skb; + swa->tso.num_sg = num_sge; + swa->tso.sgt_size = sgt_buf_size; + swa->tso.is_last_fd = total_len == 0 ? 1 : 0; + + /* Separately map the SGT buffer */ + sgt_addr = dma_map_single(dev, sgt_buf, sgt_buf_size, DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(dev, sgt_addr))) { + netdev_err(priv->net_dev, "dma_map_single(sgt_buf) failed\n"); + err = -ENOMEM; + goto err_map_sgt; + } + + /* Setup the frame descriptor */ + memset(fd, 0, sizeof(struct dpaa2_fd)); + dpaa2_fd_set_offset(fd, priv->tx_data_offset); + dpaa2_fd_set_format(fd, dpaa2_fd_sg); + dpaa2_fd_set_addr(fd, sgt_addr); + dpaa2_fd_set_len(fd, fd_len); + dpaa2_fd_set_ctrl(fd, FD_CTRL_PTA); + + *total_fds_len += fd_len; + /* Advance to the next frame descriptor */ + fd++; + index++; + } + + *num_fds = index; + + return 0; + +err_map_sgt: +err_map_data: + /* Unmap all the data S/G entries for the current FD */ + sgt = (struct dpaa2_sg_entry *)(sgt_buf + priv->tx_data_offset); + for (i = 1; i < num_sge; i++) + dma_unmap_single(dev, dpaa2_sg_get_addr(&sgt[i]), + dpaa2_sg_get_len(&sgt[i]), DMA_TO_DEVICE); + + /* Unmap the header entry */ + dma_unmap_single(dev, tso_hdr_dma, TSO_HEADER_SIZE, DMA_TO_DEVICE); +err_map_tso_hdr: + kfree(tso_hdr); +err_alloc_tso_hdr: + dpaa2_eth_sgt_recycle(priv, sgt_buf); +err_sgt_get: + /* Free all the other FDs that were already fully created */ + for (i = 0; i < index; i++) + dpaa2_eth_free_tx_fd(priv, NULL, &fd_start[i], false); + + return err; +} + +static netdev_tx_t __dpaa2_eth_tx(struct sk_buff *skb, + struct net_device *net_dev) { struct dpaa2_eth_priv *priv = netdev_priv(net_dev); - struct dpaa2_fd fd; - struct rtnl_link_stats64 *percpu_stats; + int total_enqueued = 0, retries = 0, enqueued; struct dpaa2_eth_drv_stats *percpu_extras; + struct rtnl_link_stats64 *percpu_stats; + unsigned int needed_headroom; + int num_fds = 1, max_retries; struct dpaa2_eth_fq *fq; struct netdev_queue *nq; + struct dpaa2_fd *fd; u16 queue_mapping; - unsigned int needed_headroom; - u32 fd_len; + void *swa = NULL; u8 prio = 0; int err, i; + u32 fd_len; percpu_stats = this_cpu_ptr(priv->percpu_stats); percpu_extras = this_cpu_ptr(priv->percpu_extras); + fd = (this_cpu_ptr(priv->fd))->array; - needed_headroom = dpaa2_eth_needed_headroom(priv, skb); - if (skb_headroom(skb) < needed_headroom) { - struct sk_buff *ns; - - ns = skb_realloc_headroom(skb, needed_headroom); - if (unlikely(!ns)) { - percpu_stats->tx_dropped++; - goto err_alloc_headroom; - } - percpu_extras->tx_reallocs++; - - if (skb->sk) - skb_set_owner_w(ns, skb->sk); - - dev_kfree_skb(skb); - skb = ns; - } + needed_headroom = dpaa2_eth_needed_headroom(skb); /* We'll be holding a back-reference to the skb until Tx Confirmation; * we don't want that overwritten by a concurrent Tx with a cloned skb. @@ -802,14 +1384,28 @@ static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev) } /* Setup the FD fields */ - memset(&fd, 0, sizeof(fd)); - if (skb_is_nonlinear(skb)) { - err = build_sg_fd(priv, skb, &fd); + if (skb_is_gso(skb)) { + err = dpaa2_eth_build_gso_fd(priv, skb, fd, &num_fds, &fd_len); + percpu_extras->tx_sg_frames += num_fds; + percpu_extras->tx_sg_bytes += fd_len; + percpu_extras->tx_tso_frames += num_fds; + percpu_extras->tx_tso_bytes += fd_len; + } else if (skb_is_nonlinear(skb)) { + err = dpaa2_eth_build_sg_fd(priv, skb, fd, &swa); + percpu_extras->tx_sg_frames++; + percpu_extras->tx_sg_bytes += skb->len; + fd_len = dpaa2_fd_get_len(fd); + } else if (skb_headroom(skb) < needed_headroom) { + err = dpaa2_eth_build_sg_fd_single_buf(priv, skb, fd, &swa); percpu_extras->tx_sg_frames++; percpu_extras->tx_sg_bytes += skb->len; + percpu_extras->tx_converted_sg_frames++; + percpu_extras->tx_converted_sg_bytes += skb->len; + fd_len = dpaa2_fd_get_len(fd); } else { - err = build_single_fd(priv, skb, &fd); + err = dpaa2_eth_build_single_fd(priv, skb, fd, &swa); + fd_len = dpaa2_fd_get_len(fd); } if (unlikely(err)) { @@ -817,8 +1413,12 @@ static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev) goto err_build_fd; } + if (swa && skb->cb[0]) + dpaa2_eth_enable_tx_tstamp(priv, fd, swa, skb); + /* Tracing point */ - trace_dpaa2_tx_fd(net_dev, &fd); + for (i = 0; i < num_fds; i++) + trace_dpaa2_tx_fd(net_dev, &fd[i]); /* TxConf FQ selection relies on queue id from the stack. * In case of a forwarded frame from another DPNI interface, we choose @@ -838,42 +1438,103 @@ static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev) queue_mapping %= dpaa2_eth_queue_count(priv); } fq = &priv->fq[queue_mapping]; - - fd_len = dpaa2_fd_get_len(&fd); nq = netdev_get_tx_queue(net_dev, queue_mapping); netdev_tx_sent_queue(nq, fd_len); /* Everything that happens after this enqueues might race with * the Tx confirmation callback for this frame */ - for (i = 0; i < DPAA2_ETH_ENQUEUE_RETRIES; i++) { - err = priv->enqueue(priv, fq, &fd, prio); - if (err != -EBUSY) - break; + max_retries = num_fds * DPAA2_ETH_ENQUEUE_RETRIES; + while (total_enqueued < num_fds && retries < max_retries) { + err = priv->enqueue(priv, fq, &fd[total_enqueued], + prio, num_fds - total_enqueued, &enqueued); + if (err == -EBUSY) { + retries++; + continue; + } + + total_enqueued += enqueued; } - percpu_extras->tx_portal_busy += i; + percpu_extras->tx_portal_busy += retries; + if (unlikely(err < 0)) { percpu_stats->tx_errors++; /* Clean up everything, including freeing the skb */ - free_tx_fd(priv, fq, &fd, false); + dpaa2_eth_free_tx_fd(priv, fq, fd, false); netdev_tx_completed_queue(nq, 1, fd_len); } else { - percpu_stats->tx_packets++; + percpu_stats->tx_packets += total_enqueued; percpu_stats->tx_bytes += fd_len; } return NETDEV_TX_OK; err_build_fd: -err_alloc_headroom: dev_kfree_skb(skb); return NETDEV_TX_OK; } +static void dpaa2_eth_tx_onestep_tstamp(struct work_struct *work) +{ + struct dpaa2_eth_priv *priv = container_of(work, struct dpaa2_eth_priv, + tx_onestep_tstamp); + struct sk_buff *skb; + + while (true) { + skb = skb_dequeue(&priv->tx_skbs); + if (!skb) + return; + + /* Lock just before TX one-step timestamping packet, + * and release the lock in dpaa2_eth_free_tx_fd when + * confirm the packet has been sent on hardware, or + * when clean up during transmit failure. + */ + mutex_lock(&priv->onestep_tstamp_lock); + __dpaa2_eth_tx(skb, priv->net_dev); + } +} + +static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev) +{ + struct dpaa2_eth_priv *priv = netdev_priv(net_dev); + u8 msgtype, twostep, udp; + u16 offset1, offset2; + + /* Utilize skb->cb[0] for timestamping request per skb */ + skb->cb[0] = 0; + + if ((skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP) && dpaa2_ptp) { + if (priv->tx_tstamp_type == HWTSTAMP_TX_ON) + skb->cb[0] = TX_TSTAMP; + else if (priv->tx_tstamp_type == HWTSTAMP_TX_ONESTEP_SYNC) + skb->cb[0] = TX_TSTAMP_ONESTEP_SYNC; + } + + /* TX for one-step timestamping PTP Sync packet */ + if (skb->cb[0] == TX_TSTAMP_ONESTEP_SYNC) { + if (!dpaa2_eth_ptp_parse(skb, &msgtype, &twostep, &udp, + &offset1, &offset2)) + if (msgtype == PTP_MSGTYPE_SYNC && twostep == 0) { + skb_queue_tail(&priv->tx_skbs, skb); + queue_work(priv->dpaa2_ptp_wq, + &priv->tx_onestep_tstamp); + return NETDEV_TX_OK; + } + /* Use two-step timestamping if not one-step timestamping + * PTP Sync packet + */ + skb->cb[0] = TX_TSTAMP; + } + + /* TX for other packets */ + return __dpaa2_eth_tx(skb, net_dev); +} + /* Tx confirmation frame processing routine */ static void dpaa2_eth_tx_conf(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *ch __always_unused, + struct dpaa2_eth_channel *ch, const struct dpaa2_fd *fd, struct dpaa2_eth_fq *fq) { @@ -888,10 +1549,11 @@ static void dpaa2_eth_tx_conf(struct dpaa2_eth_priv *priv, percpu_extras = this_cpu_ptr(priv->percpu_extras); percpu_extras->tx_conf_frames++; percpu_extras->tx_conf_bytes += fd_len; + ch->stats.bytes_per_cdan += fd_len; /* Check frame errors in the FD field */ fd_errors = dpaa2_fd_get_ctrl(fd) & DPAA2_FD_TX_ERR_MASK; - free_tx_fd(priv, fq, fd, true); + dpaa2_eth_free_tx_fd(priv, fq, fd, true); if (likely(!fd_errors)) return; @@ -905,7 +1567,23 @@ static void dpaa2_eth_tx_conf(struct dpaa2_eth_priv *priv, percpu_stats->tx_errors++; } -static int set_rx_csum(struct dpaa2_eth_priv *priv, bool enable) +static int dpaa2_eth_set_rx_vlan_filtering(struct dpaa2_eth_priv *priv, + bool enable) +{ + int err; + + err = dpni_enable_vlan_filter(priv->mc_io, 0, priv->mc_token, enable); + + if (err) { + netdev_err(priv->net_dev, + "dpni_enable_vlan_filter failed\n"); + return err; + } + + return 0; +} + +static int dpaa2_eth_set_rx_csum(struct dpaa2_eth_priv *priv, bool enable) { int err; @@ -928,7 +1606,7 @@ static int set_rx_csum(struct dpaa2_eth_priv *priv, bool enable) return 0; } -static int set_tx_csum(struct dpaa2_eth_priv *priv, bool enable) +static int dpaa2_eth_set_tx_csum(struct dpaa2_eth_priv *priv, bool enable) { int err; @@ -952,8 +1630,8 @@ static int set_tx_csum(struct dpaa2_eth_priv *priv, bool enable) /* Perform a single release command to add buffers * to the specified buffer pool */ -static int add_bufs(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *ch, u16 bpid) +static int dpaa2_eth_add_bufs(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, u16 bpid) { struct device *dev = priv->net_dev->dev.parent; u64 buf_array[DPAA2_ETH_BUFS_PER_CMD]; @@ -974,7 +1652,7 @@ static int add_bufs(struct dpaa2_eth_priv *priv, if (!page) goto err_alloc; - addr = dma_map_page(dev, page, 0, DPAA2_ETH_RX_BUF_SIZE, + addr = dma_map_page(dev, page, 0, priv->rx_buf_size, DMA_BIDIRECTIONAL); if (unlikely(dma_mapping_error(dev, addr))) goto err_map; @@ -982,9 +1660,9 @@ static int add_bufs(struct dpaa2_eth_priv *priv, buf_array[i] = addr; /* tracing point */ - trace_dpaa2_eth_buf_seed(priv->net_dev, - page, DPAA2_ETH_RX_BUF_RAW_SIZE, - addr, DPAA2_ETH_RX_BUF_SIZE, + trace_dpaa2_eth_buf_seed(priv->net_dev, page_address(page), + DPAA2_ETH_RX_BUF_RAW_SIZE, + addr, priv->rx_buf_size, bpid); } @@ -1001,7 +1679,7 @@ release_bufs: * not much else we can do about it */ if (err) { - free_bufs(priv, buf_array, i); + dpaa2_eth_free_bufs(priv, buf_array, i); return 0; } @@ -1019,7 +1697,7 @@ err_alloc: return 0; } -static int seed_pool(struct dpaa2_eth_priv *priv, u16 bpid) +static int dpaa2_eth_seed_pool(struct dpaa2_eth_priv *priv, u16 bpid) { int i, j; int new_count; @@ -1027,7 +1705,7 @@ static int seed_pool(struct dpaa2_eth_priv *priv, u16 bpid) for (j = 0; j < priv->num_channels; j++) { for (i = 0; i < DPAA2_ETH_NUM_BUFS; i += DPAA2_ETH_BUFS_PER_CMD) { - new_count = add_bufs(priv, priv->channel[j], bpid); + new_count = dpaa2_eth_add_bufs(priv, priv->channel[j], bpid); priv->channel[j]->buf_count += new_count; if (new_count < DPAA2_ETH_BUFS_PER_CMD) { @@ -1039,11 +1717,11 @@ static int seed_pool(struct dpaa2_eth_priv *priv, u16 bpid) return 0; } -/** +/* * Drain the specified number of buffers from the DPNI's private buffer pool. * @count must not exceeed DPAA2_ETH_BUFS_PER_CMD */ -static void drain_bufs(struct dpaa2_eth_priv *priv, int count) +static void dpaa2_eth_drain_bufs(struct dpaa2_eth_priv *priv, int count) { u64 buf_array[DPAA2_ETH_BUFS_PER_CMD]; int retries = 0; @@ -1054,22 +1732,22 @@ static void drain_bufs(struct dpaa2_eth_priv *priv, int count) buf_array, count); if (ret < 0) { if (ret == -EBUSY && - retries++ >= DPAA2_ETH_SWP_BUSY_RETRIES) + retries++ < DPAA2_ETH_SWP_BUSY_RETRIES) continue; netdev_err(priv->net_dev, "dpaa2_io_service_acquire() failed\n"); return; } - free_bufs(priv, buf_array, ret); + dpaa2_eth_free_bufs(priv, buf_array, ret); retries = 0; } while (ret); } -static void drain_pool(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_drain_pool(struct dpaa2_eth_priv *priv) { int i; - drain_bufs(priv, DPAA2_ETH_BUFS_PER_CMD); - drain_bufs(priv, 1); + dpaa2_eth_drain_bufs(priv, DPAA2_ETH_BUFS_PER_CMD); + dpaa2_eth_drain_bufs(priv, 1); for (i = 0; i < priv->num_channels; i++) priv->channel[i]->buf_count = 0; @@ -1078,9 +1756,9 @@ static void drain_pool(struct dpaa2_eth_priv *priv) /* Function is called from softirq context only, so we don't need to guard * the access to percpu count */ -static int refill_pool(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *ch, - u16 bpid) +static int dpaa2_eth_refill_pool(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *ch, + u16 bpid) { int new_count; @@ -1088,7 +1766,7 @@ static int refill_pool(struct dpaa2_eth_priv *priv, return 0; do { - new_count = add_bufs(priv, ch, bpid); + new_count = dpaa2_eth_add_bufs(priv, ch, bpid); if (unlikely(!new_count)) { /* Out of memory; abort for now, we'll try later on */ break; @@ -1102,7 +1780,23 @@ static int refill_pool(struct dpaa2_eth_priv *priv, return 0; } -static int pull_channel(struct dpaa2_eth_channel *ch) +static void dpaa2_eth_sgt_cache_drain(struct dpaa2_eth_priv *priv) +{ + struct dpaa2_eth_sgt_cache *sgt_cache; + u16 count; + int k, i; + + for_each_possible_cpu(k) { + sgt_cache = per_cpu_ptr(priv->sgt_cache, k); + count = sgt_cache->count; + + for (i = 0; i < count; i++) + skb_free_frag(sgt_cache->buf[i]); + sgt_cache->count = 0; + } +} + +static int dpaa2_eth_pull_channel(struct dpaa2_eth_channel *ch) { int err; int dequeues = -1; @@ -1138,6 +1832,7 @@ static int dpaa2_eth_poll(struct napi_struct *napi, int budget) int store_cleaned, work_done; struct list_head rx_list; int retries = 0; + u16 flowid; int err; ch = container_of(napi, struct dpaa2_eth_channel, napi); @@ -1148,18 +1843,19 @@ static int dpaa2_eth_poll(struct napi_struct *napi, int budget) ch->rx_list = &rx_list; do { - err = pull_channel(ch); + err = dpaa2_eth_pull_channel(ch); if (unlikely(err)) break; /* Refill pool if appropriate */ - refill_pool(priv, ch, priv->bpid); + dpaa2_eth_refill_pool(priv, ch, priv->bpid); - store_cleaned = consume_frames(ch, &fq); + store_cleaned = dpaa2_eth_consume_frames(ch, &fq); if (store_cleaned <= 0) break; if (fq->type == DPAA2_RX_FQ) { rx_cleaned += store_cleaned; + flowid = fq->flowid; } else { txconf_cleaned += store_cleaned; /* We have a single Tx conf FQ on this channel */ @@ -1176,6 +1872,12 @@ static int dpaa2_eth_poll(struct napi_struct *napi, int budget) } } while (store_cleaned); + /* Update NET DIM with the values for this CDAN */ + dpaa2_io_update_net_dim(ch->dpio, ch->stats.frames_per_cdan, + ch->stats.bytes_per_cdan); + ch->stats.frames_per_cdan = 0; + ch->stats.bytes_per_cdan = 0; + /* We didn't consume the entire budget, so finish napi and * re-enable data availability notifications */ @@ -1202,11 +1904,13 @@ out: if (ch->xdp.res & XDP_REDIRECT) xdp_do_flush_map(); + else if (rx_cleaned && ch->xdp.res & XDP_TX) + dpaa2_eth_xdp_tx_flush(priv, ch, &priv->fq[flowid]); return work_done; } -static void enable_ch_napi(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_enable_ch_napi(struct dpaa2_eth_priv *priv) { struct dpaa2_eth_channel *ch; int i; @@ -1217,7 +1921,7 @@ static void enable_ch_napi(struct dpaa2_eth_priv *priv) } } -static void disable_ch_napi(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_disable_ch_napi(struct dpaa2_eth_priv *priv) { struct dpaa2_eth_channel *ch; int i; @@ -1228,34 +1932,70 @@ static void disable_ch_napi(struct dpaa2_eth_priv *priv) } } -static void dpaa2_eth_set_rx_taildrop(struct dpaa2_eth_priv *priv, bool enable) +void dpaa2_eth_set_rx_taildrop(struct dpaa2_eth_priv *priv, + bool tx_pause, bool pfc) { struct dpni_taildrop td = {0}; + struct dpaa2_eth_fq *fq; int i, err; - if (priv->rx_td_enabled == enable) - return; + /* FQ taildrop: threshold is in bytes, per frame queue. Enabled if + * flow control is disabled (as it might interfere with either the + * buffer pool depletion trigger for pause frames or with the group + * congestion trigger for PFC frames) + */ + td.enable = !tx_pause; + if (priv->rx_fqtd_enabled == td.enable) + goto set_cgtd; - td.enable = enable; - td.threshold = DPAA2_ETH_TAILDROP_THRESH; + td.threshold = DPAA2_ETH_FQ_TAILDROP_THRESH; + td.units = DPNI_CONGESTION_UNIT_BYTES; for (i = 0; i < priv->num_fqs; i++) { - if (priv->fq[i].type != DPAA2_RX_FQ) + fq = &priv->fq[i]; + if (fq->type != DPAA2_RX_FQ) continue; err = dpni_set_taildrop(priv->mc_io, 0, priv->mc_token, - DPNI_CP_QUEUE, DPNI_QUEUE_RX, 0, - priv->fq[i].flowid, &td); + DPNI_CP_QUEUE, DPNI_QUEUE_RX, + fq->tc, fq->flowid, &td); if (err) { netdev_err(priv->net_dev, - "dpni_set_taildrop() failed\n"); - break; + "dpni_set_taildrop(FQ) failed\n"); + return; } } - priv->rx_td_enabled = enable; + priv->rx_fqtd_enabled = td.enable; + +set_cgtd: + /* Congestion group taildrop: threshold is in frames, per group + * of FQs belonging to the same traffic class + * Enabled if general Tx pause disabled or if PFCs are enabled + * (congestion group threhsold for PFC generation is lower than the + * CG taildrop threshold, so it won't interfere with it; we also + * want frames in non-PFC enabled traffic classes to be kept in check) + */ + td.enable = !tx_pause || pfc; + if (priv->rx_cgtd_enabled == td.enable) + return; + + td.threshold = DPAA2_ETH_CG_TAILDROP_THRESH(priv); + td.units = DPNI_CONGESTION_UNIT_FRAMES; + for (i = 0; i < dpaa2_eth_tc_count(priv); i++) { + err = dpni_set_taildrop(priv->mc_io, 0, priv->mc_token, + DPNI_CP_GROUP, DPNI_QUEUE_RX, + i, 0, &td); + if (err) { + netdev_err(priv->net_dev, + "dpni_set_taildrop(CG) failed\n"); + return; + } + } + + priv->rx_cgtd_enabled = td.enable; } -static int link_state_update(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_link_state_update(struct dpaa2_eth_priv *priv) { struct dpni_link_state state = {0}; bool tx_pause; @@ -1272,14 +2012,13 @@ static int link_state_update(struct dpaa2_eth_priv *priv) * Rx FQ taildrop configuration as well. We configure taildrop * only when pause frame generation is disabled. */ - tx_pause = !!(state.options & DPNI_LINK_OPT_PAUSE) ^ - !!(state.options & DPNI_LINK_OPT_ASYM_PAUSE); - dpaa2_eth_set_rx_taildrop(priv, !tx_pause); + tx_pause = dpaa2_eth_tx_pause_enabled(state.options); + dpaa2_eth_set_rx_taildrop(priv, tx_pause, priv->pfc_enabled); /* When we manage the MAC/PHY using phylink there is no need * to manually update the netif_carrier. */ - if (priv->mac) + if (dpaa2_eth_is_type_phy(priv)) goto out; /* Chech link state; speed / duplex changes are not treated yet */ @@ -1308,7 +2047,7 @@ static int dpaa2_eth_open(struct net_device *net_dev) struct dpaa2_eth_priv *priv = netdev_priv(net_dev); int err; - err = seed_pool(priv, priv->bpid); + err = dpaa2_eth_seed_pool(priv, priv->bpid); if (err) { /* Not much to do; the buffer pool, though not filled up, * may still contain some buffers which would enable us @@ -1318,7 +2057,7 @@ static int dpaa2_eth_open(struct net_device *net_dev) priv->dpbp_dev->obj_desc.id, priv->bpid); } - if (!priv->mac) { + if (!dpaa2_eth_is_type_phy(priv)) { /* We'll only start the txqs when the link is actually ready; * make sure we don't race against the link up notification, * which may come immediately after dpni_enable(); @@ -1332,7 +2071,7 @@ static int dpaa2_eth_open(struct net_device *net_dev) */ netif_carrier_off(net_dev); } - enable_ch_napi(priv); + dpaa2_eth_enable_ch_napi(priv); err = dpni_enable(priv->mc_io, 0, priv->mc_token); if (err < 0) { @@ -1340,30 +2079,21 @@ static int dpaa2_eth_open(struct net_device *net_dev) goto enable_err; } - if (!priv->mac) { - /* If the DPMAC object has already processed the link up - * interrupt, we have to learn the link state ourselves. - */ - err = link_state_update(priv); - if (err < 0) { - netdev_err(net_dev, "Can't update link state\n"); - goto link_state_err; - } - } else { + if (dpaa2_eth_is_type_phy(priv)) { + dpaa2_mac_start(priv->mac); phylink_start(priv->mac->phylink); } return 0; -link_state_err: enable_err: - disable_ch_napi(priv); - drain_pool(priv); + dpaa2_eth_disable_ch_napi(priv); + dpaa2_eth_drain_pool(priv); return err; } /* Total number of in-flight frames on ingress queues */ -static u32 ingress_fq_count(struct dpaa2_eth_priv *priv) +static u32 dpaa2_eth_ingress_fq_count(struct dpaa2_eth_priv *priv) { struct dpaa2_eth_fq *fq; u32 fcnt = 0, bcnt = 0, total = 0; @@ -1382,13 +2112,13 @@ static u32 ingress_fq_count(struct dpaa2_eth_priv *priv) return total; } -static void wait_for_ingress_fq_empty(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_wait_for_ingress_fq_empty(struct dpaa2_eth_priv *priv) { int retries = 10; u32 pending; do { - pending = ingress_fq_count(priv); + pending = dpaa2_eth_ingress_fq_count(priv); if (pending) msleep(100); } while (pending && --retries); @@ -1396,7 +2126,7 @@ static void wait_for_ingress_fq_empty(struct dpaa2_eth_priv *priv) #define DPNI_TX_PENDING_VER_MAJOR 7 #define DPNI_TX_PENDING_VER_MINOR 13 -static void wait_for_egress_fq_empty(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_wait_for_egress_fq_empty(struct dpaa2_eth_priv *priv) { union dpni_statistics stats; int retries = 10; @@ -1425,11 +2155,12 @@ static int dpaa2_eth_stop(struct net_device *net_dev) int dpni_enabled = 0; int retries = 10; - if (!priv->mac) { + if (dpaa2_eth_is_type_phy(priv)) { + phylink_stop(priv->mac->phylink); + dpaa2_mac_stop(priv->mac); + } else { netif_tx_stop_all_queues(net_dev); netif_carrier_off(net_dev); - } else { - phylink_stop(priv->mac->phylink); } /* On dpni_disable(), the MC firmware will: @@ -1442,7 +2173,7 @@ static int dpaa2_eth_stop(struct net_device *net_dev) * on WRIOP. After it finishes, wait until all remaining frames on Rx * and Tx conf queues are consumed on NAPI poll. */ - wait_for_egress_fq_empty(priv); + dpaa2_eth_wait_for_egress_fq_empty(priv); do { dpni_disable(priv->mc_io, 0, priv->mc_token); @@ -1458,11 +2189,14 @@ static int dpaa2_eth_stop(struct net_device *net_dev) */ } - wait_for_ingress_fq_empty(priv); - disable_ch_napi(priv); + dpaa2_eth_wait_for_ingress_fq_empty(priv); + dpaa2_eth_disable_ch_napi(priv); /* Empty the buffer pool */ - drain_pool(priv); + dpaa2_eth_drain_pool(priv); + + /* Empty the Scatter-Gather Buffer cache */ + dpaa2_eth_sgt_cache_drain(priv); return 0; } @@ -1513,8 +2247,8 @@ static void dpaa2_eth_get_stats(struct net_device *net_dev, /* Copy mac unicast addresses from @net_dev to @priv. * Its sole purpose is to make dpaa2_eth_set_rx_mode() more readable. */ -static void add_uc_hw_addr(const struct net_device *net_dev, - struct dpaa2_eth_priv *priv) +static void dpaa2_eth_add_uc_hw_addr(const struct net_device *net_dev, + struct dpaa2_eth_priv *priv) { struct netdev_hw_addr *ha; int err; @@ -1532,8 +2266,8 @@ static void add_uc_hw_addr(const struct net_device *net_dev, /* Copy mac multicast addresses from @net_dev to @priv * Its sole purpose is to make dpaa2_eth_set_rx_mode() more readable. */ -static void add_mc_hw_addr(const struct net_device *net_dev, - struct dpaa2_eth_priv *priv) +static void dpaa2_eth_add_mc_hw_addr(const struct net_device *net_dev, + struct dpaa2_eth_priv *priv) { struct netdev_hw_addr *ha; int err; @@ -1548,6 +2282,43 @@ static void add_mc_hw_addr(const struct net_device *net_dev, } } +static int dpaa2_eth_rx_add_vid(struct net_device *net_dev, + __be16 vlan_proto, u16 vid) +{ + struct dpaa2_eth_priv *priv = netdev_priv(net_dev); + int err; + + err = dpni_add_vlan_id(priv->mc_io, 0, priv->mc_token, + vid, 0, 0, 0); + + if (err) { + netdev_warn(priv->net_dev, + "Could not add the vlan id %u\n", + vid); + return err; + } + + return 0; +} + +static int dpaa2_eth_rx_kill_vid(struct net_device *net_dev, + __be16 vlan_proto, u16 vid) +{ + struct dpaa2_eth_priv *priv = netdev_priv(net_dev); + int err; + + err = dpni_remove_vlan_id(priv->mc_io, 0, priv->mc_token, vid); + + if (err) { + netdev_warn(priv->net_dev, + "Could not remove the vlan id %u\n", + vid); + return err; + } + + return 0; +} + static void dpaa2_eth_set_rx_mode(struct net_device *net_dev) { struct dpaa2_eth_priv *priv = netdev_priv(net_dev); @@ -1598,7 +2369,7 @@ static void dpaa2_eth_set_rx_mode(struct net_device *net_dev) err = dpni_clear_mac_filters(mc_io, 0, mc_token, 1, 0); if (err) netdev_warn(net_dev, "Can't clear uc filters\n"); - add_uc_hw_addr(net_dev, priv); + dpaa2_eth_add_uc_hw_addr(net_dev, priv); /* Finally, clear uc promisc and set mc promisc as requested. */ err = dpni_set_unicast_promisc(mc_io, 0, mc_token, 0); @@ -1621,8 +2392,8 @@ static void dpaa2_eth_set_rx_mode(struct net_device *net_dev) err = dpni_clear_mac_filters(mc_io, 0, mc_token, 1, 1); if (err) netdev_warn(net_dev, "Can't clear mac filters\n"); - add_mc_hw_addr(net_dev, priv); - add_uc_hw_addr(net_dev, priv); + dpaa2_eth_add_mc_hw_addr(net_dev, priv); + dpaa2_eth_add_uc_hw_addr(net_dev, priv); /* Now we can clear both ucast and mcast promisc, without risking * to drop legitimate frames anymore. @@ -1654,16 +2425,23 @@ static int dpaa2_eth_set_features(struct net_device *net_dev, bool enable; int err; + if (changed & NETIF_F_HW_VLAN_CTAG_FILTER) { + enable = !!(features & NETIF_F_HW_VLAN_CTAG_FILTER); + err = dpaa2_eth_set_rx_vlan_filtering(priv, enable); + if (err) + return err; + } + if (changed & NETIF_F_RXCSUM) { enable = !!(features & NETIF_F_RXCSUM); - err = set_rx_csum(priv, enable); + err = dpaa2_eth_set_rx_csum(priv, enable); if (err) return err; } if (changed & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM)) { enable = !!(features & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM)); - err = set_tx_csum(priv, enable); + err = dpaa2_eth_set_tx_csum(priv, enable); if (err) return err; } @@ -1676,15 +2454,17 @@ static int dpaa2_eth_ts_ioctl(struct net_device *dev, struct ifreq *rq, int cmd) struct dpaa2_eth_priv *priv = netdev_priv(dev); struct hwtstamp_config config; + if (!dpaa2_ptp) + return -EINVAL; + if (copy_from_user(&config, rq->ifr_data, sizeof(config))) return -EFAULT; switch (config.tx_type) { case HWTSTAMP_TX_OFF: - priv->tx_tstamp = false; - break; case HWTSTAMP_TX_ON: - priv->tx_tstamp = true; + case HWTSTAMP_TX_ONESTEP_SYNC: + priv->tx_tstamp_type = config.tx_type; break; default: return -ERANGE; @@ -1698,16 +2478,24 @@ static int dpaa2_eth_ts_ioctl(struct net_device *dev, struct ifreq *rq, int cmd) config.rx_filter = HWTSTAMP_FILTER_ALL; } + if (priv->tx_tstamp_type == HWTSTAMP_TX_ONESTEP_SYNC) + dpaa2_ptp_onestep_reg_update_method(priv); + return copy_to_user(rq->ifr_data, &config, sizeof(config)) ? -EFAULT : 0; } static int dpaa2_eth_ioctl(struct net_device *dev, struct ifreq *rq, int cmd) { + struct dpaa2_eth_priv *priv = netdev_priv(dev); + if (cmd == SIOCSHWTSTAMP) return dpaa2_eth_ts_ioctl(dev, rq, cmd); - return -EINVAL; + if (dpaa2_eth_is_type_phy(priv)) + return phylink_mii_ioctl(priv->mac->phylink, rq, cmd); + + return -EOPNOTSUPP; } static bool xdp_mtu_valid(struct dpaa2_eth_priv *priv, int mtu) @@ -1715,7 +2503,7 @@ static bool xdp_mtu_valid(struct dpaa2_eth_priv *priv, int mtu) int mfl, linear_mfl; mfl = DPAA2_ETH_L2_MAX_FRM(mtu); - linear_mfl = DPAA2_ETH_RX_BUF_SIZE - DPAA2_ETH_RX_HWA_SIZE - + linear_mfl = priv->rx_buf_size - DPAA2_ETH_RX_HWA_SIZE - dpaa2_eth_rx_head_room(priv) - XDP_PACKET_HEADROOM; if (mfl > linear_mfl) { @@ -1727,7 +2515,7 @@ static bool xdp_mtu_valid(struct dpaa2_eth_priv *priv, int mtu) return true; } -static int set_rx_mfl(struct dpaa2_eth_priv *priv, int mtu, bool has_xdp) +static int dpaa2_eth_set_rx_mfl(struct dpaa2_eth_priv *priv, int mtu, bool has_xdp) { int mfl, err; @@ -1761,7 +2549,7 @@ static int dpaa2_eth_change_mtu(struct net_device *dev, int new_mtu) if (!xdp_mtu_valid(priv, new_mtu)) return -EINVAL; - err = set_rx_mfl(priv, new_mtu, true); + err = dpaa2_eth_set_rx_mfl(priv, new_mtu, true); if (err) return err; @@ -1770,7 +2558,7 @@ out: return 0; } -static int update_rx_buffer_headroom(struct dpaa2_eth_priv *priv, bool has_xdp) +static int dpaa2_eth_update_rx_buffer_headroom(struct dpaa2_eth_priv *priv, bool has_xdp) { struct dpni_buffer_layout buf_layout = {0}; int err; @@ -1796,7 +2584,7 @@ static int update_rx_buffer_headroom(struct dpaa2_eth_priv *priv, bool has_xdp) return 0; } -static int setup_xdp(struct net_device *dev, struct bpf_prog *prog) +static int dpaa2_eth_setup_xdp(struct net_device *dev, struct bpf_prog *prog) { struct dpaa2_eth_priv *priv = netdev_priv(dev); struct dpaa2_eth_channel *ch; @@ -1822,10 +2610,10 @@ static int setup_xdp(struct net_device *dev, struct bpf_prog *prog) * so we are sure no old format buffers will be used from now on. */ if (need_update) { - err = set_rx_mfl(priv, dev->mtu, !!prog); + err = dpaa2_eth_set_rx_mfl(priv, dev->mtu, !!prog); if (err) goto out_err; - err = update_rx_buffer_headroom(priv, !!prog); + err = dpaa2_eth_update_rx_buffer_headroom(priv, !!prog); if (err) goto out_err; } @@ -1860,14 +2648,9 @@ out_err: static int dpaa2_eth_xdp(struct net_device *dev, struct netdev_bpf *xdp) { - struct dpaa2_eth_priv *priv = netdev_priv(dev); - switch (xdp->command) { case XDP_SETUP_PROG: - return setup_xdp(dev, xdp->prog); - case XDP_QUERY_PROG: - xdp->prog_id = priv->xdp_prog ? priv->xdp_prog->aux->id : 0; - break; + return dpaa2_eth_setup_xdp(dev, xdp->prog); default: return -EINVAL; } @@ -1875,33 +2658,25 @@ static int dpaa2_eth_xdp(struct net_device *dev, struct netdev_bpf *xdp) return 0; } -static int dpaa2_eth_xdp_xmit_frame(struct net_device *net_dev, - struct xdp_frame *xdpf) +static int dpaa2_eth_xdp_create_fd(struct net_device *net_dev, + struct xdp_frame *xdpf, + struct dpaa2_fd *fd) { - struct dpaa2_eth_priv *priv = netdev_priv(net_dev); struct device *dev = net_dev->dev.parent; - struct rtnl_link_stats64 *percpu_stats; - struct dpaa2_eth_drv_stats *percpu_extras; unsigned int needed_headroom; struct dpaa2_eth_swa *swa; - struct dpaa2_eth_fq *fq; - struct dpaa2_fd fd; void *buffer_start, *aligned_start; dma_addr_t addr; - int err, i; /* We require a minimum headroom to be able to transmit the frame. * Otherwise return an error and let the original net_device handle it */ - needed_headroom = dpaa2_eth_needed_headroom(priv, NULL); + needed_headroom = dpaa2_eth_needed_headroom(NULL); if (xdpf->headroom < needed_headroom) return -EINVAL; - percpu_stats = this_cpu_ptr(priv->percpu_stats); - percpu_extras = this_cpu_ptr(priv->percpu_extras); - /* Setup the FD fields */ - memset(&fd, 0, sizeof(fd)); + memset(fd, 0, sizeof(*fd)); /* Align FD address, if possible */ buffer_start = xdpf->data - needed_headroom; @@ -1919,32 +2694,14 @@ static int dpaa2_eth_xdp_xmit_frame(struct net_device *net_dev, addr = dma_map_single(dev, buffer_start, swa->xdp.dma_size, DMA_BIDIRECTIONAL); - if (unlikely(dma_mapping_error(dev, addr))) { - percpu_stats->tx_dropped++; + if (unlikely(dma_mapping_error(dev, addr))) return -ENOMEM; - } - dpaa2_fd_set_addr(&fd, addr); - dpaa2_fd_set_offset(&fd, xdpf->data - buffer_start); - dpaa2_fd_set_len(&fd, xdpf->len); - dpaa2_fd_set_format(&fd, dpaa2_fd_single); - dpaa2_fd_set_ctrl(&fd, FD_CTRL_PTA); - - fq = &priv->fq[smp_processor_id() % dpaa2_eth_queue_count(priv)]; - for (i = 0; i < DPAA2_ETH_ENQUEUE_RETRIES; i++) { - err = priv->enqueue(priv, fq, &fd, 0); - if (err != -EBUSY) - break; - } - percpu_extras->tx_portal_busy += i; - if (unlikely(err < 0)) { - percpu_stats->tx_errors++; - /* let the Rx device handle the cleanup */ - return err; - } - - percpu_stats->tx_packets++; - percpu_stats->tx_bytes += dpaa2_fd_get_len(&fd); + dpaa2_fd_set_addr(fd, addr); + dpaa2_fd_set_offset(fd, xdpf->data - buffer_start); + dpaa2_fd_set_len(fd, xdpf->len); + dpaa2_fd_set_format(fd, dpaa2_fd_single); + dpaa2_fd_set_ctrl(fd, FD_CTRL_PTA); return 0; } @@ -1952,8 +2709,12 @@ static int dpaa2_eth_xdp_xmit_frame(struct net_device *net_dev, static int dpaa2_eth_xdp_xmit(struct net_device *net_dev, int n, struct xdp_frame **frames, u32 flags) { - int drops = 0; - int i, err; + struct dpaa2_eth_priv *priv = netdev_priv(net_dev); + struct dpaa2_eth_xdp_fds *xdp_redirect_fds; + struct rtnl_link_stats64 *percpu_stats; + struct dpaa2_eth_fq *fq; + struct dpaa2_fd *fds; + int enqueued, i, err; if (unlikely(flags & ~XDP_XMIT_FLAGS_MASK)) return -EINVAL; @@ -1961,17 +2722,29 @@ static int dpaa2_eth_xdp_xmit(struct net_device *net_dev, int n, if (!netif_running(net_dev)) return -ENETDOWN; - for (i = 0; i < n; i++) { - struct xdp_frame *xdpf = frames[i]; + fq = &priv->fq[smp_processor_id()]; + xdp_redirect_fds = &fq->xdp_redirect_fds; + fds = xdp_redirect_fds->fds; - err = dpaa2_eth_xdp_xmit_frame(net_dev, xdpf); - if (err) { - xdp_return_frame_rx_napi(xdpf); - drops++; - } + percpu_stats = this_cpu_ptr(priv->percpu_stats); + + /* create a FD for each xdp_frame in the list received */ + for (i = 0; i < n; i++) { + err = dpaa2_eth_xdp_create_fd(net_dev, frames[i], &fds[i]); + if (err) + break; } + xdp_redirect_fds->num = i; + + /* enqueue all the frame descriptors */ + enqueued = dpaa2_eth_xdp_flush(priv, fq, xdp_redirect_fds); + + /* update statistics */ + percpu_stats->tx_packets += enqueued; + for (i = 0; i < enqueued; i++) + percpu_stats->tx_bytes += dpaa2_fd_get_len(&fds[i]); - return n - drops; + return enqueued; } static int update_xps(struct dpaa2_eth_priv *priv) @@ -2004,17 +2777,13 @@ static int update_xps(struct dpaa2_eth_priv *priv) return err; } -static int dpaa2_eth_setup_tc(struct net_device *net_dev, - enum tc_setup_type type, void *type_data) +static int dpaa2_eth_setup_mqprio(struct net_device *net_dev, + struct tc_mqprio_qopt *mqprio) { struct dpaa2_eth_priv *priv = netdev_priv(net_dev); - struct tc_mqprio_qopt *mqprio = type_data; u8 num_tc, num_queues; int i; - if (type != TC_SETUP_QDISC_MQPRIO) - return -EINVAL; - mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS; num_queues = dpaa2_eth_queue_count(priv); num_tc = mqprio->num_tc; @@ -2025,7 +2794,7 @@ static int dpaa2_eth_setup_tc(struct net_device *net_dev, if (num_tc > dpaa2_eth_tc_count(priv)) { netdev_err(net_dev, "Max %d traffic classes supported\n", dpaa2_eth_tc_count(priv)); - return -EINVAL; + return -EOPNOTSUPP; } if (!num_tc) { @@ -2046,6 +2815,60 @@ out: return 0; } +#define bps_to_mbits(rate) (div_u64((rate), 1000000) * 8) + +static int dpaa2_eth_setup_tbf(struct net_device *net_dev, struct tc_tbf_qopt_offload *p) +{ + struct tc_tbf_qopt_offload_replace_params *cfg = &p->replace_params; + struct dpaa2_eth_priv *priv = netdev_priv(net_dev); + struct dpni_tx_shaping_cfg tx_cr_shaper = { 0 }; + struct dpni_tx_shaping_cfg tx_er_shaper = { 0 }; + int err; + + if (p->command == TC_TBF_STATS) + return -EOPNOTSUPP; + + /* Only per port Tx shaping */ + if (p->parent != TC_H_ROOT) + return -EOPNOTSUPP; + + if (p->command == TC_TBF_REPLACE) { + if (cfg->max_size > DPAA2_ETH_MAX_BURST_SIZE) { + netdev_err(net_dev, "burst size cannot be greater than %d\n", + DPAA2_ETH_MAX_BURST_SIZE); + return -EINVAL; + } + + tx_cr_shaper.max_burst_size = cfg->max_size; + /* The TBF interface is in bytes/s, whereas DPAA2 expects the + * rate in Mbits/s + */ + tx_cr_shaper.rate_limit = bps_to_mbits(cfg->rate.rate_bytes_ps); + } + + err = dpni_set_tx_shaping(priv->mc_io, 0, priv->mc_token, &tx_cr_shaper, + &tx_er_shaper, 0); + if (err) { + netdev_err(net_dev, "dpni_set_tx_shaping() = %d\n", err); + return err; + } + + return 0; +} + +static int dpaa2_eth_setup_tc(struct net_device *net_dev, + enum tc_setup_type type, void *type_data) +{ + switch (type) { + case TC_SETUP_QDISC_MQPRIO: + return dpaa2_eth_setup_mqprio(net_dev, type_data); + case TC_SETUP_QDISC_TBF: + return dpaa2_eth_setup_tbf(net_dev, type_data); + default: + return -EOPNOTSUPP; + } +} + static const struct net_device_ops dpaa2_eth_ops = { .ndo_open = dpaa2_eth_open, .ndo_start_xmit = dpaa2_eth_tx, @@ -2054,14 +2877,16 @@ static const struct net_device_ops dpaa2_eth_ops = { .ndo_get_stats64 = dpaa2_eth_get_stats, .ndo_set_rx_mode = dpaa2_eth_set_rx_mode, .ndo_set_features = dpaa2_eth_set_features, - .ndo_do_ioctl = dpaa2_eth_ioctl, + .ndo_eth_ioctl = dpaa2_eth_ioctl, .ndo_change_mtu = dpaa2_eth_change_mtu, .ndo_bpf = dpaa2_eth_xdp, .ndo_xdp_xmit = dpaa2_eth_xdp_xmit, .ndo_setup_tc = dpaa2_eth_setup_tc, + .ndo_vlan_rx_add_vid = dpaa2_eth_rx_add_vid, + .ndo_vlan_rx_kill_vid = dpaa2_eth_rx_kill_vid }; -static void cdan_cb(struct dpaa2_io_notification_ctx *ctx) +static void dpaa2_eth_cdan_cb(struct dpaa2_io_notification_ctx *ctx) { struct dpaa2_eth_channel *ch; @@ -2070,11 +2895,11 @@ static void cdan_cb(struct dpaa2_io_notification_ctx *ctx) /* Update NAPI statistics */ ch->stats.cdan++; - napi_schedule_irqoff(&ch->napi); + napi_schedule(&ch->napi); } /* Allocate and configure a DPCON object */ -static struct fsl_mc_device *setup_dpcon(struct dpaa2_eth_priv *priv) +static struct fsl_mc_device *dpaa2_eth_setup_dpcon(struct dpaa2_eth_priv *priv) { struct fsl_mc_device *dpcon; struct device *dev = priv->net_dev->dev.parent; @@ -2115,19 +2940,18 @@ close: free: fsl_mc_object_free(dpcon); - return NULL; + return ERR_PTR(err); } -static void free_dpcon(struct dpaa2_eth_priv *priv, - struct fsl_mc_device *dpcon) +static void dpaa2_eth_free_dpcon(struct dpaa2_eth_priv *priv, + struct fsl_mc_device *dpcon) { dpcon_disable(priv->mc_io, 0, dpcon->mc_handle); dpcon_close(priv->mc_io, 0, dpcon->mc_handle); fsl_mc_object_free(dpcon); } -static struct dpaa2_eth_channel * -alloc_channel(struct dpaa2_eth_priv *priv) +static struct dpaa2_eth_channel *dpaa2_eth_alloc_channel(struct dpaa2_eth_priv *priv) { struct dpaa2_eth_channel *channel; struct dpcon_attr attr; @@ -2138,9 +2962,9 @@ alloc_channel(struct dpaa2_eth_priv *priv) if (!channel) return NULL; - channel->dpcon = setup_dpcon(priv); - if (IS_ERR_OR_NULL(channel->dpcon)) { - err = PTR_ERR_OR_ZERO(channel->dpcon); + channel->dpcon = dpaa2_eth_setup_dpcon(priv); + if (IS_ERR(channel->dpcon)) { + err = PTR_ERR(channel->dpcon); goto err_setup; } @@ -2158,23 +2982,23 @@ alloc_channel(struct dpaa2_eth_priv *priv) return channel; err_get_attr: - free_dpcon(priv, channel->dpcon); + dpaa2_eth_free_dpcon(priv, channel->dpcon); err_setup: kfree(channel); return ERR_PTR(err); } -static void free_channel(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_channel *channel) +static void dpaa2_eth_free_channel(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_channel *channel) { - free_dpcon(priv, channel->dpcon); + dpaa2_eth_free_dpcon(priv, channel->dpcon); kfree(channel); } /* DPIO setup: allocate and configure QBMan channels, setup core affinity * and register data availability notifications */ -static int setup_dpio(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_setup_dpio(struct dpaa2_eth_priv *priv) { struct dpaa2_io_notification_ctx *nctx; struct dpaa2_eth_channel *channel; @@ -2194,7 +3018,7 @@ static int setup_dpio(struct dpaa2_eth_priv *priv) cpumask_clear(&priv->dpio_cpumask); for_each_online_cpu(i) { /* Try to allocate a channel */ - channel = alloc_channel(priv); + channel = dpaa2_eth_alloc_channel(priv); if (IS_ERR_OR_NULL(channel)) { err = PTR_ERR_OR_ZERO(channel); if (err != -EPROBE_DEFER) @@ -2207,7 +3031,7 @@ static int setup_dpio(struct dpaa2_eth_priv *priv) nctx = &channel->nctx; nctx->is_cdan = 1; - nctx->cb = cdan_cb; + nctx->cb = dpaa2_eth_cdan_cb; nctx->id = channel->ch_id; nctx->desired_cpu = i; @@ -2255,14 +3079,14 @@ static int setup_dpio(struct dpaa2_eth_priv *priv) err_set_cdan: dpaa2_io_service_deregister(channel->dpio, nctx, dev); err_service_reg: - free_channel(priv, channel); + dpaa2_eth_free_channel(priv, channel); err_alloc_ch: if (err == -EPROBE_DEFER) { for (i = 0; i < priv->num_channels; i++) { channel = priv->channel[i]; nctx = &channel->nctx; dpaa2_io_service_deregister(channel->dpio, nctx, dev); - free_channel(priv, channel); + dpaa2_eth_free_channel(priv, channel); } priv->num_channels = 0; return err; @@ -2279,7 +3103,7 @@ err_alloc_ch: return 0; } -static void free_dpio(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_free_dpio(struct dpaa2_eth_priv *priv) { struct device *dev = priv->net_dev->dev.parent; struct dpaa2_eth_channel *ch; @@ -2289,12 +3113,12 @@ static void free_dpio(struct dpaa2_eth_priv *priv) for (i = 0; i < priv->num_channels; i++) { ch = priv->channel[i]; dpaa2_io_service_deregister(ch->dpio, &ch->nctx, dev); - free_channel(priv, ch); + dpaa2_eth_free_channel(priv, ch); } } -static struct dpaa2_eth_channel *get_affine_channel(struct dpaa2_eth_priv *priv, - int cpu) +static struct dpaa2_eth_channel *dpaa2_eth_get_affine_channel(struct dpaa2_eth_priv *priv, + int cpu) { struct device *dev = priv->net_dev->dev.parent; int i; @@ -2311,7 +3135,7 @@ static struct dpaa2_eth_channel *get_affine_channel(struct dpaa2_eth_priv *priv, return priv->channel[0]; } -static void set_fq_affinity(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_set_fq_affinity(struct dpaa2_eth_priv *priv) { struct device *dev = priv->net_dev->dev.parent; struct dpaa2_eth_fq *fq; @@ -2328,6 +3152,7 @@ static void set_fq_affinity(struct dpaa2_eth_priv *priv) fq = &priv->fq[i]; switch (fq->type) { case DPAA2_RX_FQ: + case DPAA2_RX_ERR_FQ: fq->target_cpu = rx_cpu; rx_cpu = cpumask_next(rx_cpu, &priv->dpio_cpumask); if (rx_cpu >= nr_cpu_ids) @@ -2342,15 +3167,15 @@ static void set_fq_affinity(struct dpaa2_eth_priv *priv) default: dev_err(dev, "Unknown FQ type: %d\n", fq->type); } - fq->channel = get_affine_channel(priv, fq->target_cpu); + fq->channel = dpaa2_eth_get_affine_channel(priv, fq->target_cpu); } update_xps(priv); } -static void setup_fqs(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_setup_fqs(struct dpaa2_eth_priv *priv) { - int i; + int i, j; /* We have one TxConf FQ per Tx flow. * The number of Tx and Rx queues is the same. @@ -2362,18 +3187,25 @@ static void setup_fqs(struct dpaa2_eth_priv *priv) priv->fq[priv->num_fqs++].flowid = (u16)i; } - for (i = 0; i < dpaa2_eth_queue_count(priv); i++) { - priv->fq[priv->num_fqs].type = DPAA2_RX_FQ; - priv->fq[priv->num_fqs].consume = dpaa2_eth_rx; - priv->fq[priv->num_fqs++].flowid = (u16)i; + for (j = 0; j < dpaa2_eth_tc_count(priv); j++) { + for (i = 0; i < dpaa2_eth_queue_count(priv); i++) { + priv->fq[priv->num_fqs].type = DPAA2_RX_FQ; + priv->fq[priv->num_fqs].consume = dpaa2_eth_rx; + priv->fq[priv->num_fqs].tc = (u8)j; + priv->fq[priv->num_fqs++].flowid = (u16)i; + } } + /* We have exactly one Rx error queue per DPNI */ + priv->fq[priv->num_fqs].type = DPAA2_RX_ERR_FQ; + priv->fq[priv->num_fqs++].consume = dpaa2_eth_rx_err; + /* For each FQ, decide on which core to process incoming frames */ - set_fq_affinity(priv); + dpaa2_eth_set_fq_affinity(priv); } /* Allocate and configure one buffer pool for each interface */ -static int setup_dpbp(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_setup_dpbp(struct dpaa2_eth_priv *priv) { int err; struct fsl_mc_device *dpbp_dev; @@ -2432,15 +3264,15 @@ err_open: return err; } -static void free_dpbp(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_free_dpbp(struct dpaa2_eth_priv *priv) { - drain_pool(priv); + dpaa2_eth_drain_pool(priv); dpbp_disable(priv->mc_io, 0, priv->dpbp_dev->mc_handle); dpbp_close(priv->mc_io, 0, priv->dpbp_dev->mc_handle); fsl_mc_object_free(priv->dpbp_dev); } -static int set_buffer_layout(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_set_buffer_layout(struct dpaa2_eth_priv *priv) { struct device *dev = priv->net_dev->dev.parent; struct dpni_buffer_layout buf_layout = {0}; @@ -2457,11 +3289,18 @@ static int set_buffer_layout(struct dpaa2_eth_priv *priv) else rx_buf_align = DPAA2_ETH_RX_BUF_ALIGN; + /* We need to ensure that the buffer size seen by WRIOP is a multiple + * of 64 or 256 bytes depending on the WRIOP version. + */ + priv->rx_buf_size = ALIGN_DOWN(DPAA2_ETH_RX_BUF_SIZE, rx_buf_align); + /* tx buffer */ buf_layout.private_data_size = DPAA2_ETH_SWA_SIZE; buf_layout.pass_timestamp = true; + buf_layout.pass_frame_status = true; buf_layout.options = DPNI_BUF_LAYOUT_OPT_PRIVATE_DATA_SIZE | - DPNI_BUF_LAYOUT_OPT_TIMESTAMP; + DPNI_BUF_LAYOUT_OPT_TIMESTAMP | + DPNI_BUF_LAYOUT_OPT_FRAME_STATUS; err = dpni_set_buffer_layout(priv->mc_io, 0, priv->mc_token, DPNI_QUEUE_TX, &buf_layout); if (err) { @@ -2470,7 +3309,8 @@ static int set_buffer_layout(struct dpaa2_eth_priv *priv) } /* tx-confirm buffer */ - buf_layout.options = DPNI_BUF_LAYOUT_OPT_TIMESTAMP; + buf_layout.options = DPNI_BUF_LAYOUT_OPT_TIMESTAMP | + DPNI_BUF_LAYOUT_OPT_FRAME_STATUS; err = dpni_set_buffer_layout(priv->mc_io, 0, priv->mc_token, DPNI_QUEUE_TX_CONFIRM, &buf_layout); if (err) { @@ -2518,31 +3358,50 @@ static int set_buffer_layout(struct dpaa2_eth_priv *priv) static inline int dpaa2_eth_enqueue_qd(struct dpaa2_eth_priv *priv, struct dpaa2_eth_fq *fq, - struct dpaa2_fd *fd, u8 prio) + struct dpaa2_fd *fd, u8 prio, + u32 num_frames __always_unused, + int *frames_enqueued) { - return dpaa2_io_service_enqueue_qd(fq->channel->dpio, - priv->tx_qdid, prio, - fq->tx_qdbin, fd); + int err; + + err = dpaa2_io_service_enqueue_qd(fq->channel->dpio, + priv->tx_qdid, prio, + fq->tx_qdbin, fd); + if (!err && frames_enqueued) + *frames_enqueued = 1; + return err; } -static inline int dpaa2_eth_enqueue_fq(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_fq *fq, - struct dpaa2_fd *fd, u8 prio) +static inline int dpaa2_eth_enqueue_fq_multiple(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_fq *fq, + struct dpaa2_fd *fd, + u8 prio, u32 num_frames, + int *frames_enqueued) { - return dpaa2_io_service_enqueue_fq(fq->channel->dpio, - fq->tx_fqid[prio], fd); + int err; + + err = dpaa2_io_service_enqueue_multiple_fq(fq->channel->dpio, + fq->tx_fqid[prio], + fd, num_frames); + + if (err == 0) + return -EBUSY; + + if (frames_enqueued) + *frames_enqueued = err; + return 0; } -static void set_enqueue_mode(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_set_enqueue_mode(struct dpaa2_eth_priv *priv) { if (dpaa2_eth_cmp_dpni_ver(priv, DPNI_ENQUEUE_FQID_VER_MAJOR, DPNI_ENQUEUE_FQID_VER_MINOR) < 0) priv->enqueue = dpaa2_eth_enqueue_qd; else - priv->enqueue = dpaa2_eth_enqueue_fq; + priv->enqueue = dpaa2_eth_enqueue_fq_multiple; } -static int set_pause(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_set_pause(struct dpaa2_eth_priv *priv) { struct device *dev = priv->net_dev->dev.parent; struct dpni_link_cfg link_cfg = {0}; @@ -2569,7 +3428,7 @@ static int set_pause(struct dpaa2_eth_priv *priv) return 0; } -static void update_tx_fqids(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_update_tx_fqids(struct dpaa2_eth_priv *priv) { struct dpni_queue_id qid = {0}; struct dpaa2_eth_fq *fq; @@ -2600,7 +3459,7 @@ static void update_tx_fqids(struct dpaa2_eth_priv *priv) } } - priv->enqueue = dpaa2_eth_enqueue_fq; + priv->enqueue = dpaa2_eth_enqueue_fq_multiple; return; @@ -2610,8 +3469,120 @@ out_err: priv->enqueue = dpaa2_eth_enqueue_qd; } +/* Configure ingress classification based on VLAN PCP */ +static int dpaa2_eth_set_vlan_qos(struct dpaa2_eth_priv *priv) +{ + struct device *dev = priv->net_dev->dev.parent; + struct dpkg_profile_cfg kg_cfg = {0}; + struct dpni_qos_tbl_cfg qos_cfg = {0}; + struct dpni_rule_cfg key_params; + void *dma_mem, *key, *mask; + u8 key_size = 2; /* VLAN TCI field */ + int i, pcp, err; + + /* VLAN-based classification only makes sense if we have multiple + * traffic classes. + * Also, we need to extract just the 3-bit PCP field from the VLAN + * header and we can only do that by using a mask + */ + if (dpaa2_eth_tc_count(priv) == 1 || !dpaa2_eth_fs_mask_enabled(priv)) { + dev_dbg(dev, "VLAN-based QoS classification not supported\n"); + return -EOPNOTSUPP; + } + + dma_mem = kzalloc(DPAA2_CLASSIFIER_DMA_SIZE, GFP_KERNEL); + if (!dma_mem) + return -ENOMEM; + + kg_cfg.num_extracts = 1; + kg_cfg.extracts[0].type = DPKG_EXTRACT_FROM_HDR; + kg_cfg.extracts[0].extract.from_hdr.prot = NET_PROT_VLAN; + kg_cfg.extracts[0].extract.from_hdr.type = DPKG_FULL_FIELD; + kg_cfg.extracts[0].extract.from_hdr.field = NH_FLD_VLAN_TCI; + + err = dpni_prepare_key_cfg(&kg_cfg, dma_mem); + if (err) { + dev_err(dev, "dpni_prepare_key_cfg failed\n"); + goto out_free_tbl; + } + + /* set QoS table */ + qos_cfg.default_tc = 0; + qos_cfg.discard_on_miss = 0; + qos_cfg.key_cfg_iova = dma_map_single(dev, dma_mem, + DPAA2_CLASSIFIER_DMA_SIZE, + DMA_TO_DEVICE); + if (dma_mapping_error(dev, qos_cfg.key_cfg_iova)) { + dev_err(dev, "QoS table DMA mapping failed\n"); + err = -ENOMEM; + goto out_free_tbl; + } + + err = dpni_set_qos_table(priv->mc_io, 0, priv->mc_token, &qos_cfg); + if (err) { + dev_err(dev, "dpni_set_qos_table failed\n"); + goto out_unmap_tbl; + } + + /* Add QoS table entries */ + key = kzalloc(key_size * 2, GFP_KERNEL); + if (!key) { + err = -ENOMEM; + goto out_unmap_tbl; + } + mask = key + key_size; + *(__be16 *)mask = cpu_to_be16(VLAN_PRIO_MASK); + + key_params.key_iova = dma_map_single(dev, key, key_size * 2, + DMA_TO_DEVICE); + if (dma_mapping_error(dev, key_params.key_iova)) { + dev_err(dev, "Qos table entry DMA mapping failed\n"); + err = -ENOMEM; + goto out_free_key; + } + + key_params.mask_iova = key_params.key_iova + key_size; + key_params.key_size = key_size; + + /* We add rules for PCP-based distribution starting with highest + * priority (VLAN PCP = 7). If this DPNI doesn't have enough traffic + * classes to accommodate all priority levels, the lowest ones end up + * on TC 0 which was configured as default + */ + for (i = dpaa2_eth_tc_count(priv) - 1, pcp = 7; i >= 0; i--, pcp--) { + *(__be16 *)key = cpu_to_be16(pcp << VLAN_PRIO_SHIFT); + dma_sync_single_for_device(dev, key_params.key_iova, + key_size * 2, DMA_TO_DEVICE); + + err = dpni_add_qos_entry(priv->mc_io, 0, priv->mc_token, + &key_params, i, i); + if (err) { + dev_err(dev, "dpni_add_qos_entry failed\n"); + dpni_clear_qos_table(priv->mc_io, 0, priv->mc_token); + goto out_unmap_key; + } + } + + priv->vlan_cls_enabled = true; + + /* Table and key memory is not persistent, clean everything up after + * configuration is finished + */ +out_unmap_key: + dma_unmap_single(dev, key_params.key_iova, key_size * 2, DMA_TO_DEVICE); +out_free_key: + kfree(key); +out_unmap_tbl: + dma_unmap_single(dev, qos_cfg.key_cfg_iova, DPAA2_CLASSIFIER_DMA_SIZE, + DMA_TO_DEVICE); +out_free_tbl: + kfree(dma_mem); + + return err; +} + /* Configure the DPNI object this interface is associated with */ -static int setup_dpni(struct fsl_mc_device *ls_dev) +static int dpaa2_eth_setup_dpni(struct fsl_mc_device *ls_dev) { struct device *dev = &ls_dev->dev; struct dpaa2_eth_priv *priv; @@ -2659,23 +3630,30 @@ static int setup_dpni(struct fsl_mc_device *ls_dev) goto close; } - err = set_buffer_layout(priv); + err = dpaa2_eth_set_buffer_layout(priv); if (err) goto close; - set_enqueue_mode(priv); + dpaa2_eth_set_enqueue_mode(priv); /* Enable pause frame support */ if (dpaa2_eth_has_pause_support(priv)) { - err = set_pause(priv); + err = dpaa2_eth_set_pause(priv); if (err) goto close; } - priv->cls_rules = devm_kzalloc(dev, sizeof(struct dpaa2_eth_cls_rule) * - dpaa2_eth_fs_count(priv), GFP_KERNEL); - if (!priv->cls_rules) + err = dpaa2_eth_set_vlan_qos(priv); + if (err && err != -EOPNOTSUPP) + goto close; + + priv->cls_rules = devm_kcalloc(dev, dpaa2_eth_fs_count(priv), + sizeof(struct dpaa2_eth_cls_rule), + GFP_KERNEL); + if (!priv->cls_rules) { + err = -ENOMEM; goto close; + } return 0; @@ -2685,7 +3663,7 @@ close: return err; } -static void free_dpni(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_free_dpni(struct dpaa2_eth_priv *priv) { int err; @@ -2697,8 +3675,8 @@ static void free_dpni(struct dpaa2_eth_priv *priv) dpni_close(priv->mc_io, 0, priv->mc_token); } -static int setup_rx_flow(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_fq *fq) +static int dpaa2_eth_setup_rx_flow(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_fq *fq) { struct device *dev = priv->net_dev->dev.parent; struct dpni_queue queue; @@ -2706,7 +3684,7 @@ static int setup_rx_flow(struct dpaa2_eth_priv *priv, int err; err = dpni_get_queue(priv->mc_io, 0, priv->mc_token, - DPNI_QUEUE_RX, 0, fq->flowid, &queue, &qid); + DPNI_QUEUE_RX, fq->tc, fq->flowid, &queue, &qid); if (err) { dev_err(dev, "dpni_get_queue(RX) failed\n"); return err; @@ -2719,7 +3697,7 @@ static int setup_rx_flow(struct dpaa2_eth_priv *priv, queue.destination.priority = 1; queue.user_context = (u64)(uintptr_t)fq; err = dpni_set_queue(priv->mc_io, 0, priv->mc_token, - DPNI_QUEUE_RX, 0, fq->flowid, + DPNI_QUEUE_RX, fq->tc, fq->flowid, DPNI_QUEUE_OPT_USER_CTX | DPNI_QUEUE_OPT_DEST, &queue); if (err) { @@ -2728,8 +3706,12 @@ static int setup_rx_flow(struct dpaa2_eth_priv *priv, } /* xdp_rxq setup */ + /* only once for each channel */ + if (fq->tc > 0) + return 0; + err = xdp_rxq_info_reg(&fq->channel->xdp_rxq, priv->net_dev, - fq->flowid); + fq->flowid, 0); if (err) { dev_err(dev, "xdp_rxq_info_reg failed\n"); return err; @@ -2745,8 +3727,8 @@ static int setup_rx_flow(struct dpaa2_eth_priv *priv, return 0; } -static int setup_tx_flow(struct dpaa2_eth_priv *priv, - struct dpaa2_eth_fq *fq) +static int dpaa2_eth_setup_tx_flow(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_fq *fq) { struct device *dev = priv->net_dev->dev.parent; struct dpni_queue queue; @@ -2793,6 +3775,38 @@ static int setup_tx_flow(struct dpaa2_eth_priv *priv, return 0; } +static int setup_rx_err_flow(struct dpaa2_eth_priv *priv, + struct dpaa2_eth_fq *fq) +{ + struct device *dev = priv->net_dev->dev.parent; + struct dpni_queue q = { { 0 } }; + struct dpni_queue_id qid; + u8 q_opt = DPNI_QUEUE_OPT_USER_CTX | DPNI_QUEUE_OPT_DEST; + int err; + + err = dpni_get_queue(priv->mc_io, 0, priv->mc_token, + DPNI_QUEUE_RX_ERR, 0, 0, &q, &qid); + if (err) { + dev_err(dev, "dpni_get_queue() failed (%d)\n", err); + return err; + } + + fq->fqid = qid.fqid; + + q.destination.id = fq->channel->dpcon_id; + q.destination.type = DPNI_DEST_DPCON; + q.destination.priority = 1; + q.user_context = (u64)(uintptr_t)fq; + err = dpni_set_queue(priv->mc_io, 0, priv->mc_token, + DPNI_QUEUE_RX_ERR, 0, 0, q_opt, &q); + if (err) { + dev_err(dev, "dpni_set_queue() failed (%d)\n", err); + return err; + } + + return 0; +} + /* Supported header fields for Rx hash distribution key */ static const struct dpaa2_eth_dist_fields dist_fields[] = { { @@ -2861,11 +3875,11 @@ static const struct dpaa2_eth_dist_fields dist_fields[] = { }; /* Configure the Rx hash key using the legacy API */ -static int config_legacy_hash_key(struct dpaa2_eth_priv *priv, dma_addr_t key) +static int dpaa2_eth_config_legacy_hash_key(struct dpaa2_eth_priv *priv, dma_addr_t key) { struct device *dev = priv->net_dev->dev.parent; struct dpni_rx_tc_dist_cfg dist_cfg; - int err; + int i, err = 0; memset(&dist_cfg, 0, sizeof(dist_cfg)); @@ -2873,19 +3887,24 @@ static int config_legacy_hash_key(struct dpaa2_eth_priv *priv, dma_addr_t key) dist_cfg.dist_size = dpaa2_eth_queue_count(priv); dist_cfg.dist_mode = DPNI_DIST_MODE_HASH; - err = dpni_set_rx_tc_dist(priv->mc_io, 0, priv->mc_token, 0, &dist_cfg); - if (err) - dev_err(dev, "dpni_set_rx_tc_dist failed\n"); + for (i = 0; i < dpaa2_eth_tc_count(priv); i++) { + err = dpni_set_rx_tc_dist(priv->mc_io, 0, priv->mc_token, + i, &dist_cfg); + if (err) { + dev_err(dev, "dpni_set_rx_tc_dist failed\n"); + break; + } + } return err; } /* Configure the Rx hash key using the new API */ -static int config_hash_key(struct dpaa2_eth_priv *priv, dma_addr_t key) +static int dpaa2_eth_config_hash_key(struct dpaa2_eth_priv *priv, dma_addr_t key) { struct device *dev = priv->net_dev->dev.parent; struct dpni_rx_dist_cfg dist_cfg; - int err; + int i, err = 0; memset(&dist_cfg, 0, sizeof(dist_cfg)); @@ -2893,19 +3912,31 @@ static int config_hash_key(struct dpaa2_eth_priv *priv, dma_addr_t key) dist_cfg.dist_size = dpaa2_eth_queue_count(priv); dist_cfg.enable = 1; - err = dpni_set_rx_hash_dist(priv->mc_io, 0, priv->mc_token, &dist_cfg); - if (err) - dev_err(dev, "dpni_set_rx_hash_dist failed\n"); + for (i = 0; i < dpaa2_eth_tc_count(priv); i++) { + dist_cfg.tc = i; + err = dpni_set_rx_hash_dist(priv->mc_io, 0, priv->mc_token, + &dist_cfg); + if (err) { + dev_err(dev, "dpni_set_rx_hash_dist failed\n"); + break; + } + + /* If the flow steering / hashing key is shared between all + * traffic classes, install it just once + */ + if (priv->dpni_attrs.options & DPNI_OPT_SHARED_FS) + break; + } return err; } /* Configure the Rx flow classification key */ -static int config_cls_key(struct dpaa2_eth_priv *priv, dma_addr_t key) +static int dpaa2_eth_config_cls_key(struct dpaa2_eth_priv *priv, dma_addr_t key) { struct device *dev = priv->net_dev->dev.parent; struct dpni_rx_dist_cfg dist_cfg; - int err; + int i, err = 0; memset(&dist_cfg, 0, sizeof(dist_cfg)); @@ -2913,9 +3944,21 @@ static int config_cls_key(struct dpaa2_eth_priv *priv, dma_addr_t key) dist_cfg.dist_size = dpaa2_eth_queue_count(priv); dist_cfg.enable = 1; - err = dpni_set_rx_fs_dist(priv->mc_io, 0, priv->mc_token, &dist_cfg); - if (err) - dev_err(dev, "dpni_set_rx_fs_dist failed\n"); + for (i = 0; i < dpaa2_eth_tc_count(priv); i++) { + dist_cfg.tc = i; + err = dpni_set_rx_fs_dist(priv->mc_io, 0, priv->mc_token, + &dist_cfg); + if (err) { + dev_err(dev, "dpni_set_rx_fs_dist failed\n"); + break; + } + + /* If the flow steering / hashing key is shared between all + * traffic classes, install it just once + */ + if (priv->dpni_attrs.options & DPNI_OPT_SHARED_FS) + break; + } return err; } @@ -3030,11 +4073,11 @@ static int dpaa2_eth_set_dist_key(struct net_device *net_dev, if (type == DPAA2_ETH_RX_DIST_HASH) { if (dpaa2_eth_has_legacy_dist(priv)) - err = config_legacy_hash_key(priv, key_iova); + err = dpaa2_eth_config_legacy_hash_key(priv, key_iova); else - err = config_hash_key(priv, key_iova); + err = dpaa2_eth_config_hash_key(priv, key_iova); } else { - err = config_cls_key(priv, key_iova); + err = dpaa2_eth_config_cls_key(priv, key_iova); } dma_unmap_single(dev, key_iova, DPAA2_CLASSIFIER_DMA_SIZE, @@ -3109,7 +4152,7 @@ out: /* Bind the DPNI to its needed objects and resources: buffer pool, DPIOs, * frame queues and channels */ -static int bind_dpni(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_bind_dpni(struct dpaa2_eth_priv *priv) { struct net_device *net_dev = priv->net_dev; struct device *dev = net_dev->dev.parent; @@ -3121,7 +4164,7 @@ static int bind_dpni(struct dpaa2_eth_priv *priv) pools_params.num_dpbp = 1; pools_params.pools[0].dpbp_id = priv->dpbp_dev->obj_desc.id; pools_params.pools[0].backup_pool = 0; - pools_params.pools[0].buffer_size = DPAA2_ETH_RX_BUF_SIZE; + pools_params.pools[0].buffer_size = priv->rx_buf_size; err = dpni_set_pools(priv->mc_io, 0, priv->mc_token, &pools_params); if (err) { dev_err(dev, "dpni_set_pools() failed\n"); @@ -3157,10 +4200,13 @@ static int bind_dpni(struct dpaa2_eth_priv *priv) for (i = 0; i < priv->num_fqs; i++) { switch (priv->fq[i].type) { case DPAA2_RX_FQ: - err = setup_rx_flow(priv, &priv->fq[i]); + err = dpaa2_eth_setup_rx_flow(priv, &priv->fq[i]); break; case DPAA2_TX_CONF_FQ: - err = setup_tx_flow(priv, &priv->fq[i]); + err = dpaa2_eth_setup_tx_flow(priv, &priv->fq[i]); + break; + case DPAA2_RX_ERR_FQ: + err = setup_rx_err_flow(priv, &priv->fq[i]); break; default: dev_err(dev, "Invalid FQ type %d\n", priv->fq[i].type); @@ -3181,7 +4227,7 @@ static int bind_dpni(struct dpaa2_eth_priv *priv) } /* Allocate rings for storing incoming frame descriptors */ -static int alloc_rings(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_alloc_rings(struct dpaa2_eth_priv *priv) { struct net_device *net_dev = priv->net_dev; struct device *dev = net_dev->dev.parent; @@ -3208,7 +4254,7 @@ err_ring: return -ENOMEM; } -static void free_rings(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_free_rings(struct dpaa2_eth_priv *priv) { int i; @@ -3216,7 +4262,7 @@ static void free_rings(struct dpaa2_eth_priv *priv) dpaa2_io_store_destroy(priv->channel[i]->store); } -static int set_mac_addr(struct dpaa2_eth_priv *priv) +static int dpaa2_eth_set_mac_addr(struct dpaa2_eth_priv *priv) { struct net_device *net_dev = priv->net_dev; struct device *dev = net_dev->dev.parent; @@ -3250,7 +4296,7 @@ static int set_mac_addr(struct dpaa2_eth_priv *priv) return err; } } - memcpy(net_dev->dev_addr, mac_addr, net_dev->addr_len); + eth_hw_addr_set(net_dev, mac_addr); } else if (is_zero_ether_addr(dpni_mac_addr)) { /* No MAC address configured, fill in net_dev->dev_addr * with a random one @@ -3275,13 +4321,13 @@ static int set_mac_addr(struct dpaa2_eth_priv *priv) /* NET_ADDR_PERM is default, all we have to do is * fill in the device addr. */ - memcpy(net_dev->dev_addr, dpni_mac_addr, net_dev->addr_len); + eth_hw_addr_set(net_dev, dpni_mac_addr); } return 0; } -static int netdev_init(struct net_device *net_dev) +static int dpaa2_eth_netdev_init(struct net_device *net_dev) { struct device *dev = net_dev->dev.parent; struct dpaa2_eth_priv *priv = netdev_priv(net_dev); @@ -3294,7 +4340,7 @@ static int netdev_init(struct net_device *net_dev) net_dev->netdev_ops = &dpaa2_eth_ops; net_dev->ethtool_ops = &dpaa2_ethtool_ops; - err = set_mac_addr(priv); + err = dpaa2_eth_set_mac_addr(priv); if (err) return err; @@ -3328,6 +4374,8 @@ static int netdev_init(struct net_device *net_dev) return err; } + dpaa2_eth_detect_features(priv); + /* Capabilities listing */ supported |= IFF_LIVE_ADDR_CHANGE; @@ -3343,19 +4391,23 @@ static int netdev_init(struct net_device *net_dev) net_dev->features = NETIF_F_RXCSUM | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_SG | NETIF_F_HIGHDMA | - NETIF_F_LLTX; + NETIF_F_LLTX | NETIF_F_HW_TC | NETIF_F_TSO; + net_dev->gso_max_segs = DPAA2_ETH_ENQUEUE_MAX_FDS; net_dev->hw_features = net_dev->features; + if (priv->dpni_attrs.vlan_filter_entries) + net_dev->hw_features |= NETIF_F_HW_VLAN_CTAG_FILTER; + return 0; } -static int poll_link_state(void *arg) +static int dpaa2_eth_poll_link_state(void *arg) { struct dpaa2_eth_priv *priv = (struct dpaa2_eth_priv *)arg; int err; while (!kthread_should_stop()) { - err = link_state_update(priv); + err = dpaa2_eth_link_state_update(priv); if (unlikely(err)) return err; @@ -3372,11 +4424,12 @@ static int dpaa2_eth_connect_mac(struct dpaa2_eth_priv *priv) int err; dpni_dev = to_fsl_mc_device(priv->net_dev->dev.parent); - dpmac_dev = fsl_mc_get_endpoint(dpni_dev); - if (IS_ERR(dpmac_dev) || dpmac_dev->dev.type != &fsl_mc_bus_dpmac_type) - return 0; + dpmac_dev = fsl_mc_get_endpoint(dpni_dev, 0); - if (dpaa2_mac_is_type_fixed(dpmac_dev, priv->mc_io)) + if (PTR_ERR(dpmac_dev) == -EPROBE_DEFER) + return PTR_ERR(dpmac_dev); + + if (IS_ERR(dpmac_dev) || dpmac_dev->dev.type != &fsl_mc_bus_dpmac_type) return 0; mac = kzalloc(sizeof(struct dpaa2_mac), GFP_KERNEL); @@ -3387,23 +4440,39 @@ static int dpaa2_eth_connect_mac(struct dpaa2_eth_priv *priv) mac->mc_io = priv->mc_io; mac->net_dev = priv->net_dev; - err = dpaa2_mac_connect(mac); - if (err) { - netdev_err(priv->net_dev, "Error connecting to the MAC endpoint\n"); - kfree(mac); - return err; - } + err = dpaa2_mac_open(mac); + if (err) + goto err_free_mac; priv->mac = mac; + if (dpaa2_eth_is_type_phy(priv)) { + err = dpaa2_mac_connect(mac); + if (err && err != -EPROBE_DEFER) + netdev_err(priv->net_dev, "Error connecting to the MAC endpoint: %pe", + ERR_PTR(err)); + if (err) + goto err_close_mac; + } + return 0; + +err_close_mac: + dpaa2_mac_close(mac); + priv->mac = NULL; +err_free_mac: + kfree(mac); + return err; } static void dpaa2_eth_disconnect_mac(struct dpaa2_eth_priv *priv) { - if (!priv->mac) + if (dpaa2_eth_is_type_phy(priv)) + dpaa2_mac_disconnect(priv->mac); + + if (!dpaa2_eth_has_mac(priv)) return; - dpaa2_mac_disconnect(priv->mac); + dpaa2_mac_close(priv->mac); kfree(priv->mac); priv->mac = NULL; } @@ -3425,14 +4494,14 @@ static irqreturn_t dpni_irq0_handler_thread(int irq_num, void *arg) } if (status & DPNI_IRQ_EVENT_LINK_CHANGED) - link_state_update(netdev_priv(net_dev)); + dpaa2_eth_link_state_update(netdev_priv(net_dev)); if (status & DPNI_IRQ_EVENT_ENDPOINT_CHANGED) { - set_mac_addr(netdev_priv(net_dev)); - update_tx_fqids(priv); + dpaa2_eth_set_mac_addr(netdev_priv(net_dev)); + dpaa2_eth_update_tx_fqids(priv); rtnl_lock(); - if (priv->mac) + if (dpaa2_eth_has_mac(priv)) dpaa2_eth_disconnect_mac(priv); else dpaa2_eth_connect_mac(priv); @@ -3442,7 +4511,7 @@ static irqreturn_t dpni_irq0_handler_thread(int irq_num, void *arg) return IRQ_HANDLED; } -static int setup_irqs(struct fsl_mc_device *ls_dev) +static int dpaa2_eth_setup_irqs(struct fsl_mc_device *ls_dev) { int err = 0; struct fsl_mc_device_irq *irq; @@ -3454,7 +4523,7 @@ static int setup_irqs(struct fsl_mc_device *ls_dev) } irq = ls_dev->irqs[0]; - err = devm_request_threaded_irq(&ls_dev->dev, irq->msi_desc->irq, + err = devm_request_threaded_irq(&ls_dev->dev, irq->virq, NULL, dpni_irq0_handler_thread, IRQF_NO_SUSPEND | IRQF_ONESHOT, dev_name(&ls_dev->dev), &ls_dev->dev); @@ -3481,14 +4550,14 @@ static int setup_irqs(struct fsl_mc_device *ls_dev) return 0; free_irq: - devm_free_irq(&ls_dev->dev, irq->msi_desc->irq, &ls_dev->dev); + devm_free_irq(&ls_dev->dev, irq->virq, &ls_dev->dev); free_mc_irq: fsl_mc_free_irqs(ls_dev); return err; } -static void add_ch_napi(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_add_ch_napi(struct dpaa2_eth_priv *priv) { int i; struct dpaa2_eth_channel *ch; @@ -3496,12 +4565,11 @@ static void add_ch_napi(struct dpaa2_eth_priv *priv) for (i = 0; i < priv->num_channels; i++) { ch = priv->channel[i]; /* NAPI weight *MUST* be a multiple of DPAA2_ETH_STORE_SIZE */ - netif_napi_add(priv->net_dev, &ch->napi, dpaa2_eth_poll, - NAPI_POLL_WEIGHT); + netif_napi_add(priv->net_dev, &ch->napi, dpaa2_eth_poll); } } -static void del_ch_napi(struct dpaa2_eth_priv *priv) +static void dpaa2_eth_del_ch_napi(struct dpaa2_eth_priv *priv) { int i; struct dpaa2_eth_channel *ch; @@ -3536,6 +4604,21 @@ static int dpaa2_eth_probe(struct fsl_mc_device *dpni_dev) priv->iommu_domain = iommu_get_domain_for_dev(dev); + priv->tx_tstamp_type = HWTSTAMP_TX_OFF; + priv->rx_tstamp = false; + + priv->dpaa2_ptp_wq = alloc_workqueue("dpaa2_ptp_wq", 0, 0); + if (!priv->dpaa2_ptp_wq) { + err = -ENOMEM; + goto err_wq_alloc; + } + + INIT_WORK(&priv->tx_onestep_tstamp, dpaa2_eth_tx_onestep_tstamp); + mutex_init(&priv->onestep_tstamp_lock); + skb_queue_head_init(&priv->tx_skbs); + + priv->rx_copybreak = DPAA2_ETH_DEFAULT_COPYBREAK; + /* Obtain a MC portal */ err = fsl_mc_portal_allocate(dpni_dev, FSL_MC_IO_ATOMIC_CONTEXT_PORTAL, &priv->mc_io); @@ -3548,26 +4631,26 @@ static int dpaa2_eth_probe(struct fsl_mc_device *dpni_dev) } /* MC objects initialization and configuration */ - err = setup_dpni(dpni_dev); + err = dpaa2_eth_setup_dpni(dpni_dev); if (err) goto err_dpni_setup; - err = setup_dpio(priv); + err = dpaa2_eth_setup_dpio(priv); if (err) goto err_dpio_setup; - setup_fqs(priv); + dpaa2_eth_setup_fqs(priv); - err = setup_dpbp(priv); + err = dpaa2_eth_setup_dpbp(priv); if (err) goto err_dpbp_setup; - err = bind_dpni(priv); + err = dpaa2_eth_bind_dpni(priv); if (err) goto err_bind; /* Add a NAPI context for each channel */ - add_ch_napi(priv); + dpaa2_eth_add_ch_napi(priv); /* Percpu statistics */ priv->percpu_stats = alloc_percpu(*priv->percpu_stats); @@ -3583,28 +4666,51 @@ static int dpaa2_eth_probe(struct fsl_mc_device *dpni_dev) goto err_alloc_percpu_extras; } - err = netdev_init(net_dev); + priv->sgt_cache = alloc_percpu(*priv->sgt_cache); + if (!priv->sgt_cache) { + dev_err(dev, "alloc_percpu(sgt_cache) failed\n"); + err = -ENOMEM; + goto err_alloc_sgt_cache; + } + + priv->fd = alloc_percpu(*priv->fd); + if (!priv->fd) { + dev_err(dev, "alloc_percpu(fds) failed\n"); + err = -ENOMEM; + goto err_alloc_fds; + } + + err = dpaa2_eth_netdev_init(net_dev); if (err) goto err_netdev_init; /* Configure checksum offload based on current interface flags */ - err = set_rx_csum(priv, !!(net_dev->features & NETIF_F_RXCSUM)); + err = dpaa2_eth_set_rx_csum(priv, !!(net_dev->features & NETIF_F_RXCSUM)); if (err) goto err_csum; - err = set_tx_csum(priv, !!(net_dev->features & - (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM))); + err = dpaa2_eth_set_tx_csum(priv, + !!(net_dev->features & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM))); if (err) goto err_csum; - err = alloc_rings(priv); + err = dpaa2_eth_alloc_rings(priv); if (err) goto err_alloc_rings; - err = setup_irqs(dpni_dev); +#ifdef CONFIG_FSL_DPAA2_ETH_DCB + if (dpaa2_eth_has_pause_support(priv) && priv->vlan_cls_enabled) { + priv->dcbx_mode = DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_IEEE; + net_dev->dcbnl_ops = &dpaa2_eth_dcbnl_ops; + } else { + dev_dbg(dev, "PFC not supported\n"); + } +#endif + + err = dpaa2_eth_setup_irqs(dpni_dev); if (err) { netdev_warn(net_dev, "Failed to set link interrupt, fall back to polling\n"); - priv->poll_thread = kthread_run(poll_link_state, priv, + priv->poll_thread = kthread_run(dpaa2_eth_poll_link_state, priv, "%s_poll_link", net_dev->name); if (IS_ERR(priv->poll_thread)) { dev_err(dev, "Error starting polling thread\n"); @@ -3617,6 +4723,18 @@ static int dpaa2_eth_probe(struct fsl_mc_device *dpni_dev) if (err) goto err_connect_mac; + err = dpaa2_eth_dl_alloc(priv); + if (err) + goto err_dl_register; + + err = dpaa2_eth_dl_traps_register(priv); + if (err) + goto err_dl_trap_register; + + err = dpaa2_eth_dl_port_add(priv); + if (err) + goto err_dl_port_add; + err = register_netdev(net_dev); if (err < 0) { dev_err(dev, "register_netdev() failed\n"); @@ -3627,10 +4745,17 @@ static int dpaa2_eth_probe(struct fsl_mc_device *dpni_dev) dpaa2_dbg_add(priv); #endif + dpaa2_eth_dl_register(priv); dev_info(dev, "Probed interface %s\n", net_dev->name); return 0; err_netdev_reg: + dpaa2_eth_dl_port_del(priv); +err_dl_port_add: + dpaa2_eth_dl_traps_unregister(priv); +err_dl_trap_register: + dpaa2_eth_dl_free(priv); +err_dl_register: dpaa2_eth_disconnect_mac(priv); err_connect_mac: if (priv->do_link_poll) @@ -3638,24 +4763,30 @@ err_connect_mac: else fsl_mc_free_irqs(dpni_dev); err_poll_thread: - free_rings(priv); + dpaa2_eth_free_rings(priv); err_alloc_rings: err_csum: err_netdev_init: + free_percpu(priv->fd); +err_alloc_fds: + free_percpu(priv->sgt_cache); +err_alloc_sgt_cache: free_percpu(priv->percpu_extras); err_alloc_percpu_extras: free_percpu(priv->percpu_stats); err_alloc_percpu_stats: - del_ch_napi(priv); + dpaa2_eth_del_ch_napi(priv); err_bind: - free_dpbp(priv); + dpaa2_eth_free_dpbp(priv); err_dpbp_setup: - free_dpio(priv); + dpaa2_eth_free_dpio(priv); err_dpio_setup: - free_dpni(priv); + dpaa2_eth_free_dpni(priv); err_dpni_setup: fsl_mc_portal_free(priv->mc_io); err_portal_alloc: + destroy_workqueue(priv->dpaa2_ptp_wq); +err_wq_alloc: dev_set_drvdata(dev, NULL); free_netdev(net_dev); @@ -3672,35 +4803,47 @@ static int dpaa2_eth_remove(struct fsl_mc_device *ls_dev) net_dev = dev_get_drvdata(dev); priv = netdev_priv(net_dev); + dpaa2_eth_dl_unregister(priv); + #ifdef CONFIG_DEBUG_FS dpaa2_dbg_remove(priv); #endif + + unregister_netdev(net_dev); rtnl_lock(); dpaa2_eth_disconnect_mac(priv); rtnl_unlock(); - unregister_netdev(net_dev); + dpaa2_eth_dl_port_del(priv); + dpaa2_eth_dl_traps_unregister(priv); + dpaa2_eth_dl_free(priv); if (priv->do_link_poll) kthread_stop(priv->poll_thread); else fsl_mc_free_irqs(ls_dev); - free_rings(priv); + dpaa2_eth_free_rings(priv); + free_percpu(priv->fd); + free_percpu(priv->sgt_cache); free_percpu(priv->percpu_stats); free_percpu(priv->percpu_extras); - del_ch_napi(priv); - free_dpbp(priv); - free_dpio(priv); - free_dpni(priv); + dpaa2_eth_del_ch_napi(priv); + dpaa2_eth_free_dpbp(priv); + dpaa2_eth_free_dpio(priv); + dpaa2_eth_free_dpni(priv); + if (priv->onestep_reg_base) + iounmap(priv->onestep_reg_base); fsl_mc_portal_free(priv->mc_io); - free_netdev(net_dev); + destroy_workqueue(priv->dpaa2_ptp_wq); dev_dbg(net_dev->dev.parent, "Removed interface %s\n", net_dev->name); + free_netdev(net_dev); + return 0; } |