diff options
Diffstat (limited to 'drivers/net/ethernet/qlogic/qede/qede_main.c')
| -rw-r--r-- | drivers/net/ethernet/qlogic/qede/qede_main.c | 1503 |
1 files changed, 937 insertions, 566 deletions
diff --git a/drivers/net/ethernet/qlogic/qede/qede_main.c b/drivers/net/ethernet/qlogic/qede/qede_main.c index 7def29aaf65c..aecdd1c5c0ea 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_main.c +++ b/drivers/net/ethernet/qlogic/qede/qede_main.c @@ -94,11 +94,26 @@ static int qede_probe(struct pci_dev *pdev, const struct pci_device_id *id); #define TX_TIMEOUT (5 * HZ) +/* Utilize last protocol index for XDP */ +#define XDP_PI 11 + static void qede_remove(struct pci_dev *pdev); -static int qede_alloc_rx_buffer(struct qede_dev *edev, - struct qede_rx_queue *rxq); +static void qede_shutdown(struct pci_dev *pdev); static void qede_link_update(void *dev, struct qed_link_output *link); +/* The qede lock is used to protect driver state change and driver flows that + * are not reentrant. + */ +void __qede_lock(struct qede_dev *edev) +{ + mutex_lock(&edev->qede_lock); +} + +void __qede_unlock(struct qede_dev *edev) +{ + mutex_unlock(&edev->qede_lock); +} + #ifdef CONFIG_QED_SRIOV static int qede_set_vf_vlan(struct net_device *ndev, int vf, u16 vlan, u8 qos, __be16 vlan_proto) @@ -166,15 +181,20 @@ static struct pci_driver qede_pci_driver = { .id_table = qede_pci_tbl, .probe = qede_probe, .remove = qede_remove, + .shutdown = qede_shutdown, #ifdef CONFIG_QED_SRIOV .sriov_configure = qede_sriov_configure, #endif }; -static void qede_force_mac(void *dev, u8 *mac) +static void qede_force_mac(void *dev, u8 *mac, bool forced) { struct qede_dev *edev = dev; + /* MAC hints take effect only if we haven't set one already */ + if (is_valid_ether_addr(edev->ndev->dev_addr) && !forced) + return; + ether_addr_copy(edev->ndev->dev_addr, mac); ether_addr_copy(edev->primary_mac, mac); } @@ -284,12 +304,12 @@ static int qede_free_tx_pkt(struct qede_dev *edev, struct qede_tx_queue *txq, int *len) { u16 idx = txq->sw_tx_cons & NUM_TX_BDS_MAX; - struct sk_buff *skb = txq->sw_tx_ring[idx].skb; + struct sk_buff *skb = txq->sw_tx_ring.skbs[idx].skb; struct eth_tx_1st_bd *first_bd; struct eth_tx_bd *tx_data_bd; int bds_consumed = 0; int nbds; - bool data_split = txq->sw_tx_ring[idx].flags & QEDE_TSO_SPLIT_BD; + bool data_split = txq->sw_tx_ring.skbs[idx].flags & QEDE_TSO_SPLIT_BD; int i, split_bd_len = 0; if (unlikely(!skb)) { @@ -329,20 +349,19 @@ static int qede_free_tx_pkt(struct qede_dev *edev, /* Free skb */ dev_kfree_skb_any(skb); - txq->sw_tx_ring[idx].skb = NULL; - txq->sw_tx_ring[idx].flags = 0; + txq->sw_tx_ring.skbs[idx].skb = NULL; + txq->sw_tx_ring.skbs[idx].flags = 0; return 0; } /* Unmap the data and free skb when mapping failed during start_xmit */ -static void qede_free_failed_tx_pkt(struct qede_dev *edev, - struct qede_tx_queue *txq, +static void qede_free_failed_tx_pkt(struct qede_tx_queue *txq, struct eth_tx_1st_bd *first_bd, int nbd, bool data_split) { u16 idx = txq->sw_tx_prod & NUM_TX_BDS_MAX; - struct sk_buff *skb = txq->sw_tx_ring[idx].skb; + struct sk_buff *skb = txq->sw_tx_ring.skbs[idx].skb; struct eth_tx_bd *tx_data_bd; int i, split_bd_len = 0; @@ -359,7 +378,7 @@ static void qede_free_failed_tx_pkt(struct qede_dev *edev, nbd--; } - dma_unmap_single(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), + dma_unmap_single(txq->dev, BD_UNMAP_ADDR(first_bd), BD_UNMAP_LEN(first_bd) + split_bd_len, DMA_TO_DEVICE); /* Unmap the data of the skb frags */ @@ -367,7 +386,7 @@ static void qede_free_failed_tx_pkt(struct qede_dev *edev, tx_data_bd = (struct eth_tx_bd *) qed_chain_produce(&txq->tx_pbl); if (tx_data_bd->nbytes) - dma_unmap_page(&edev->pdev->dev, + dma_unmap_page(txq->dev, BD_UNMAP_ADDR(tx_data_bd), BD_UNMAP_LEN(tx_data_bd), DMA_TO_DEVICE); } @@ -378,12 +397,11 @@ static void qede_free_failed_tx_pkt(struct qede_dev *edev, /* Free skb */ dev_kfree_skb_any(skb); - txq->sw_tx_ring[idx].skb = NULL; - txq->sw_tx_ring[idx].flags = 0; + txq->sw_tx_ring.skbs[idx].skb = NULL; + txq->sw_tx_ring.skbs[idx].flags = 0; } -static u32 qede_xmit_type(struct qede_dev *edev, - struct sk_buff *skb, int *ipv6_ext) +static u32 qede_xmit_type(struct sk_buff *skb, int *ipv6_ext) { u32 rc = XMIT_L4_CSUM; __be16 l3_proto; @@ -396,8 +414,19 @@ static u32 qede_xmit_type(struct qede_dev *edev, (ipv6_hdr(skb)->nexthdr == NEXTHDR_IPV6)) *ipv6_ext = 1; - if (skb->encapsulation) + if (skb->encapsulation) { rc |= XMIT_ENC; + if (skb_is_gso(skb)) { + unsigned short gso_type = skb_shinfo(skb)->gso_type; + + if ((gso_type & SKB_GSO_UDP_TUNNEL_CSUM) || + (gso_type & SKB_GSO_GRE_CSUM)) + rc |= XMIT_ENC_GSO_L4_CSUM; + + rc |= XMIT_LSO; + return rc; + } + } if (skb_is_gso(skb)) rc |= XMIT_LSO; @@ -439,18 +468,16 @@ static void qede_set_params_for_ipv6_ext(struct sk_buff *skb, second_bd->data.bitfields2 = cpu_to_le16(bd2_bits2); } -static int map_frag_to_bd(struct qede_dev *edev, +static int map_frag_to_bd(struct qede_tx_queue *txq, skb_frag_t *frag, struct eth_tx_bd *bd) { dma_addr_t mapping; /* Map skb non-linear frag data for DMA */ - mapping = skb_frag_dma_map(&edev->pdev->dev, frag, 0, + mapping = skb_frag_dma_map(txq->dev, frag, 0, skb_frag_size(frag), DMA_TO_DEVICE); - if (unlikely(dma_mapping_error(&edev->pdev->dev, mapping))) { - DP_NOTICE(edev, "Unable to map frag - dropping packet\n"); + if (unlikely(dma_mapping_error(txq->dev, mapping))) return -ENOMEM; - } /* Setup the data pointer of the frag data */ BD_SET_UNMAP_ADDR_LEN(bd, mapping, skb_frag_size(frag)); @@ -470,8 +497,7 @@ static u16 qede_get_skb_hlen(struct sk_buff *skb, bool is_encap_pkt) /* +2 for 1st BD for headers and 2nd BD for headlen (if required) */ #if ((MAX_SKB_FRAGS + 2) > ETH_TX_MAX_BDS_PER_NON_LSO_PACKET) -static bool qede_pkt_req_lin(struct qede_dev *edev, struct sk_buff *skb, - u8 xmit_type) +static bool qede_pkt_req_lin(struct sk_buff *skb, u8 xmit_type) { int allowed_frags = ETH_TX_MAX_BDS_PER_NON_LSO_PACKET - 1; @@ -507,6 +533,47 @@ static inline void qede_update_tx_producer(struct qede_tx_queue *txq) mmiowb(); } +static int qede_xdp_xmit(struct qede_dev *edev, struct qede_fastpath *fp, + struct sw_rx_data *metadata, u16 padding, u16 length) +{ + struct qede_tx_queue *txq = fp->xdp_tx; + u16 idx = txq->sw_tx_prod & NUM_TX_BDS_MAX; + struct eth_tx_1st_bd *first_bd; + + if (!qed_chain_get_elem_left(&txq->tx_pbl)) { + txq->stopped_cnt++; + return -ENOMEM; + } + + first_bd = (struct eth_tx_1st_bd *)qed_chain_produce(&txq->tx_pbl); + + memset(first_bd, 0, sizeof(*first_bd)); + first_bd->data.bd_flags.bitfields = + BIT(ETH_TX_1ST_BD_FLAGS_START_BD_SHIFT); + first_bd->data.bitfields |= + (length & ETH_TX_DATA_1ST_BD_PKT_LEN_MASK) << + ETH_TX_DATA_1ST_BD_PKT_LEN_SHIFT; + first_bd->data.nbds = 1; + + /* We can safely ignore the offset, as it's 0 for XDP */ + BD_SET_UNMAP_ADDR_LEN(first_bd, metadata->mapping + padding, length); + + /* Synchronize the buffer back to device, as program [probably] + * has changed it. + */ + dma_sync_single_for_device(&edev->pdev->dev, + metadata->mapping + padding, + length, PCI_DMA_TODEVICE); + + txq->sw_tx_ring.pages[idx] = metadata->data; + txq->sw_tx_prod++; + + /* Mark the fastpath for future XDP doorbell */ + fp->xdp_xmit = 1; + + return 0; +} + /* Main transmit function */ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, struct net_device *ndev) @@ -530,15 +597,15 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, /* Get tx-queue context and netdev index */ txq_index = skb_get_queue_mapping(skb); WARN_ON(txq_index >= QEDE_TSS_COUNT(edev)); - txq = QEDE_TX_QUEUE(edev, txq_index); + txq = edev->fp_array[edev->fp_num_rx + txq_index].txq; netdev_txq = netdev_get_tx_queue(ndev, txq_index); WARN_ON(qed_chain_get_elem_left(&txq->tx_pbl) < (MAX_SKB_FRAGS + 1)); - xmit_type = qede_xmit_type(edev, skb, &ipv6_ext); + xmit_type = qede_xmit_type(skb, &ipv6_ext); #if ((MAX_SKB_FRAGS + 2) > ETH_TX_MAX_BDS_PER_NON_LSO_PACKET) - if (qede_pkt_req_lin(edev, skb, xmit_type)) { + if (qede_pkt_req_lin(skb, xmit_type)) { if (skb_linearize(skb)) { DP_NOTICE(edev, "SKB linearization failed - silently dropping this SKB\n"); @@ -550,7 +617,7 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, /* Fill the entry in the SW ring and the BDs in the FW ring */ idx = txq->sw_tx_prod & NUM_TX_BDS_MAX; - txq->sw_tx_ring[idx].skb = skb; + txq->sw_tx_ring.skbs[idx].skb = skb; first_bd = (struct eth_tx_1st_bd *) qed_chain_produce(&txq->tx_pbl); memset(first_bd, 0, sizeof(*first_bd)); @@ -558,11 +625,11 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, 1 << ETH_TX_1ST_BD_FLAGS_START_BD_SHIFT; /* Map skb linear data for DMA and set in the first BD */ - mapping = dma_map_single(&edev->pdev->dev, skb->data, + mapping = dma_map_single(txq->dev, skb->data, skb_headlen(skb), DMA_TO_DEVICE); - if (unlikely(dma_mapping_error(&edev->pdev->dev, mapping))) { + if (unlikely(dma_mapping_error(txq->dev, mapping))) { DP_NOTICE(edev, "SKB mapping failed\n"); - qede_free_failed_tx_pkt(edev, txq, first_bd, 0, false); + qede_free_failed_tx_pkt(txq, first_bd, 0, false); qede_update_tx_producer(txq); return NETDEV_TX_OK; } @@ -633,6 +700,12 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, if (unlikely(xmit_type & XMIT_ENC)) { first_bd->data.bd_flags.bitfields |= 1 << ETH_TX_1ST_BD_FLAGS_TUNN_IP_CSUM_SHIFT; + + if (xmit_type & XMIT_ENC_GSO_L4_CSUM) { + u8 tmp = ETH_TX_1ST_BD_FLAGS_TUNN_L4_CSUM_SHIFT; + + first_bd->data.bd_flags.bitfields |= 1 << tmp; + } hlen = qede_get_skb_hlen(skb, true); } else { first_bd->data.bd_flags.bitfields |= @@ -664,7 +737,7 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, /* this marks the BD as one that has no * individual mapping */ - txq->sw_tx_ring[idx].flags |= QEDE_TSO_SPLIT_BD; + txq->sw_tx_ring.skbs[idx].flags |= QEDE_TSO_SPLIT_BD; first_bd->nbytes = cpu_to_le16(hlen); @@ -680,12 +753,11 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, /* Handle fragmented skb */ /* special handle for frags inside 2nd and 3rd bds.. */ while (tx_data_bd && frag_idx < skb_shinfo(skb)->nr_frags) { - rc = map_frag_to_bd(edev, + rc = map_frag_to_bd(txq, &skb_shinfo(skb)->frags[frag_idx], tx_data_bd); if (rc) { - qede_free_failed_tx_pkt(edev, txq, first_bd, nbd, - data_split); + qede_free_failed_tx_pkt(txq, first_bd, nbd, data_split); qede_update_tx_producer(txq); return NETDEV_TX_OK; } @@ -705,12 +777,11 @@ static netdev_tx_t qede_start_xmit(struct sk_buff *skb, memset(tx_data_bd, 0, sizeof(*tx_data_bd)); - rc = map_frag_to_bd(edev, + rc = map_frag_to_bd(txq, &skb_shinfo(skb)->frags[frag_idx], tx_data_bd); if (rc) { - qede_free_failed_tx_pkt(edev, txq, first_bd, nbd, - data_split); + qede_free_failed_tx_pkt(txq, first_bd, nbd, data_split); qede_update_tx_producer(txq); return NETDEV_TX_OK; } @@ -775,6 +846,27 @@ int qede_txq_has_work(struct qede_tx_queue *txq) return hw_bd_cons != qed_chain_get_cons_idx(&txq->tx_pbl); } +static void qede_xdp_tx_int(struct qede_dev *edev, struct qede_tx_queue *txq) +{ + struct eth_tx_1st_bd *bd; + u16 hw_bd_cons; + + hw_bd_cons = le16_to_cpu(*txq->hw_cons_ptr); + barrier(); + + while (hw_bd_cons != qed_chain_get_cons_idx(&txq->tx_pbl)) { + bd = (struct eth_tx_1st_bd *)qed_chain_consume(&txq->tx_pbl); + + dma_unmap_single(&edev->pdev->dev, BD_UNMAP_ADDR(bd), + PAGE_SIZE, DMA_BIDIRECTIONAL); + __free_page(txq->sw_tx_ring.pages[txq->sw_tx_cons & + NUM_TX_BDS_MAX]); + + txq->sw_tx_cons++; + txq->xmit_pkts++; + } +} + static int qede_tx_int(struct qede_dev *edev, struct qede_tx_queue *txq) { struct netdev_queue *netdev_txq; @@ -858,16 +950,6 @@ bool qede_has_rx_work(struct qede_rx_queue *rxq) return hw_comp_cons != sw_comp_cons; } -static bool qede_has_tx_work(struct qede_fastpath *fp) -{ - u8 tc; - - for (tc = 0; tc < fp->edev->num_tc; tc++) - if (qede_txq_has_work(&fp->txqs[tc])) - return true; - return false; -} - static inline void qede_rx_bd_ring_consume(struct qede_rx_queue *rxq) { qed_chain_consume(&rxq->rx_bd_ring); @@ -877,8 +959,7 @@ static inline void qede_rx_bd_ring_consume(struct qede_rx_queue *rxq) /* This function reuses the buffer(from an offset) from * consumer index to producer index in the bd ring */ -static inline void qede_reuse_page(struct qede_dev *edev, - struct qede_rx_queue *rxq, +static inline void qede_reuse_page(struct qede_rx_queue *rxq, struct sw_rx_data *curr_cons) { struct eth_rx_bd *rx_bd_prod = qed_chain_produce(&rxq->rx_bd_ring); @@ -900,27 +981,62 @@ static inline void qede_reuse_page(struct qede_dev *edev, /* In case of allocation failures reuse buffers * from consumer index to produce buffers for firmware */ -void qede_recycle_rx_bd_ring(struct qede_rx_queue *rxq, - struct qede_dev *edev, u8 count) +void qede_recycle_rx_bd_ring(struct qede_rx_queue *rxq, u8 count) { struct sw_rx_data *curr_cons; for (; count > 0; count--) { curr_cons = &rxq->sw_rx_ring[rxq->sw_rx_cons & NUM_RX_BDS_MAX]; - qede_reuse_page(edev, rxq, curr_cons); + qede_reuse_page(rxq, curr_cons); qede_rx_bd_ring_consume(rxq); } } -static inline int qede_realloc_rx_buffer(struct qede_dev *edev, - struct qede_rx_queue *rxq, +static int qede_alloc_rx_buffer(struct qede_rx_queue *rxq) +{ + struct sw_rx_data *sw_rx_data; + struct eth_rx_bd *rx_bd; + dma_addr_t mapping; + struct page *data; + + data = alloc_pages(GFP_ATOMIC, 0); + if (unlikely(!data)) + return -ENOMEM; + + /* Map the entire page as it would be used + * for multiple RX buffer segment size mapping. + */ + mapping = dma_map_page(rxq->dev, data, 0, + PAGE_SIZE, rxq->data_direction); + if (unlikely(dma_mapping_error(rxq->dev, mapping))) { + __free_page(data); + return -ENOMEM; + } + + sw_rx_data = &rxq->sw_rx_ring[rxq->sw_rx_prod & NUM_RX_BDS_MAX]; + sw_rx_data->page_offset = 0; + sw_rx_data->data = data; + sw_rx_data->mapping = mapping; + + /* Advance PROD and get BD pointer */ + rx_bd = (struct eth_rx_bd *)qed_chain_produce(&rxq->rx_bd_ring); + WARN_ON(!rx_bd); + rx_bd->addr.hi = cpu_to_le32(upper_32_bits(mapping)); + rx_bd->addr.lo = cpu_to_le32(lower_32_bits(mapping)); + + rxq->sw_rx_prod++; + + return 0; +} + +static inline int qede_realloc_rx_buffer(struct qede_rx_queue *rxq, struct sw_rx_data *curr_cons) { /* Move to the next segment in the page */ curr_cons->page_offset += rxq->rx_buf_seg_size; if (curr_cons->page_offset == PAGE_SIZE) { - if (unlikely(qede_alloc_rx_buffer(edev, rxq))) { + if (unlikely(qede_alloc_rx_buffer(rxq))) { /* Since we failed to allocate new buffer * current buffer can be used again. */ @@ -929,15 +1045,15 @@ static inline int qede_realloc_rx_buffer(struct qede_dev *edev, return -ENOMEM; } - dma_unmap_page(&edev->pdev->dev, curr_cons->mapping, - PAGE_SIZE, DMA_FROM_DEVICE); + dma_unmap_page(rxq->dev, curr_cons->mapping, + PAGE_SIZE, rxq->data_direction); } else { /* Increment refcount of the page as we don't want * network stack to take the ownership of the page * which can be recycled multiple times by the driver. */ page_ref_inc(curr_cons->data); - qede_reuse_page(edev, rxq, curr_cons); + qede_reuse_page(rxq, curr_cons); } return 0; @@ -971,22 +1087,20 @@ void qede_update_rx_prod(struct qede_dev *edev, struct qede_rx_queue *rxq) mmiowb(); } -static u32 qede_get_rxhash(struct qede_dev *edev, - u8 bitfields, - __le32 rss_hash, enum pkt_hash_types *rxhash_type) +static void qede_get_rxhash(struct sk_buff *skb, u8 bitfields, __le32 rss_hash) { + enum pkt_hash_types hash_type = PKT_HASH_TYPE_NONE; enum rss_hash_type htype; + u32 hash = 0; htype = GET_FIELD(bitfields, ETH_FAST_PATH_RX_REG_CQE_RSS_HASH_TYPE); - - if ((edev->ndev->features & NETIF_F_RXHASH) && htype) { - *rxhash_type = ((htype == RSS_HASH_TYPE_IPV4) || - (htype == RSS_HASH_TYPE_IPV6)) ? - PKT_HASH_TYPE_L3 : PKT_HASH_TYPE_L4; - return le32_to_cpu(rss_hash); + if (htype) { + hash_type = ((htype == RSS_HASH_TYPE_IPV4) || + (htype == RSS_HASH_TYPE_IPV6)) ? + PKT_HASH_TYPE_L3 : PKT_HASH_TYPE_L4; + hash = le32_to_cpu(rss_hash); } - *rxhash_type = PKT_HASH_TYPE_NONE; - return 0; + skb_set_hash(skb, hash, hash_type); } static void qede_set_skb_csum(struct sk_buff *skb, u8 csum_flag) @@ -1002,12 +1116,14 @@ static void qede_set_skb_csum(struct sk_buff *skb, u8 csum_flag) static inline void qede_skb_receive(struct qede_dev *edev, struct qede_fastpath *fp, + struct qede_rx_queue *rxq, struct sk_buff *skb, u16 vlan_tag) { if (vlan_tag) __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), vlan_tag); napi_gro_receive(&fp->napi, skb); + fp->rxq->rcv_pkts++; } static void qede_set_gro_params(struct qede_dev *edev, @@ -1035,7 +1151,7 @@ static int qede_fill_frag_skb(struct qede_dev *edev, struct qede_agg_info *tpa_info = &rxq->tpa_info[tpa_agg_index]; struct sk_buff *skb = tpa_info->skb; - if (unlikely(tpa_info->agg_state != QEDE_AGG_STATE_START)) + if (unlikely(tpa_info->state != QEDE_AGG_STATE_START)) goto out; /* Add one frag and update the appropriate fields in the skb */ @@ -1043,7 +1159,7 @@ static int qede_fill_frag_skb(struct qede_dev *edev, current_bd->data, current_bd->page_offset, len_on_bd); - if (unlikely(qede_realloc_rx_buffer(edev, rxq, current_bd))) { + if (unlikely(qede_realloc_rx_buffer(rxq, current_bd))) { /* Incr page ref count to reuse on allocation failure * so that it doesn't get freed while freeing SKB. */ @@ -1061,8 +1177,9 @@ static int qede_fill_frag_skb(struct qede_dev *edev, return 0; out: - tpa_info->agg_state = QEDE_AGG_STATE_ERROR; - qede_recycle_rx_bd_ring(rxq, edev, 1); + tpa_info->state = QEDE_AGG_STATE_ERROR; + qede_recycle_rx_bd_ring(rxq, 1); + return -ENOMEM; } @@ -1073,12 +1190,10 @@ static void qede_tpa_start(struct qede_dev *edev, struct qede_agg_info *tpa_info = &rxq->tpa_info[cqe->tpa_agg_index]; struct eth_rx_bd *rx_bd_cons = qed_chain_consume(&rxq->rx_bd_ring); struct eth_rx_bd *rx_bd_prod = qed_chain_produce(&rxq->rx_bd_ring); - struct sw_rx_data *replace_buf = &tpa_info->replace_buf; - dma_addr_t mapping = tpa_info->replace_buf_mapping; + struct sw_rx_data *replace_buf = &tpa_info->buffer; + dma_addr_t mapping = tpa_info->buffer_mapping; struct sw_rx_data *sw_rx_data_cons; struct sw_rx_data *sw_rx_data_prod; - enum pkt_hash_types rxhash_type; - u32 rxhash; sw_rx_data_cons = &rxq->sw_rx_ring[rxq->sw_rx_cons & NUM_RX_BDS_MAX]; sw_rx_data_prod = &rxq->sw_rx_ring[rxq->sw_rx_prod & NUM_RX_BDS_MAX]; @@ -1099,11 +1214,11 @@ static void qede_tpa_start(struct qede_dev *edev, /* move partial skb from cons to pool (don't unmap yet) * save mapping, incase we drop the packet later on. */ - tpa_info->start_buf = *sw_rx_data_cons; + tpa_info->buffer = *sw_rx_data_cons; mapping = HILO_U64(le32_to_cpu(rx_bd_cons->addr.hi), le32_to_cpu(rx_bd_cons->addr.lo)); - tpa_info->start_buf_mapping = mapping; + tpa_info->buffer_mapping = mapping; rxq->sw_rx_cons++; /* set tpa state to start only if we are able to allocate skb @@ -1114,27 +1229,27 @@ static void qede_tpa_start(struct qede_dev *edev, le16_to_cpu(cqe->len_on_first_bd)); if (unlikely(!tpa_info->skb)) { DP_NOTICE(edev, "Failed to allocate SKB for gro\n"); - tpa_info->agg_state = QEDE_AGG_STATE_ERROR; + tpa_info->state = QEDE_AGG_STATE_ERROR; goto cons_buf; } - skb_put(tpa_info->skb, le16_to_cpu(cqe->len_on_first_bd)); - memcpy(&tpa_info->start_cqe, cqe, sizeof(tpa_info->start_cqe)); - /* Start filling in the aggregation info */ + skb_put(tpa_info->skb, le16_to_cpu(cqe->len_on_first_bd)); tpa_info->frag_id = 0; - tpa_info->agg_state = QEDE_AGG_STATE_START; + tpa_info->state = QEDE_AGG_STATE_START; - rxhash = qede_get_rxhash(edev, cqe->bitfields, - cqe->rss_hash, &rxhash_type); - skb_set_hash(tpa_info->skb, rxhash, rxhash_type); + /* Store some information from first CQE */ + tpa_info->start_cqe_placement_offset = cqe->placement_offset; + tpa_info->start_cqe_bd_len = le16_to_cpu(cqe->len_on_first_bd); if ((le16_to_cpu(cqe->pars_flags.flags) >> PARSING_AND_ERR_FLAGS_TAG8021QEXIST_SHIFT) & - PARSING_AND_ERR_FLAGS_TAG8021QEXIST_MASK) + PARSING_AND_ERR_FLAGS_TAG8021QEXIST_MASK) tpa_info->vlan_tag = le16_to_cpu(cqe->vlan_tag); else tpa_info->vlan_tag = 0; + qede_get_rxhash(tpa_info->skb, cqe->bitfields, cqe->rss_hash); + /* This is needed in order to enable forwarding support */ qede_set_gro_params(edev, tpa_info->skb, cqe); @@ -1146,7 +1261,7 @@ cons_buf: /* We still need to handle bd_len_list to consume buffers */ if (unlikely(cqe->ext_bd_len_list[1])) { DP_ERR(edev, "Unlikely - got a TPA aggregation with more than one ext_bd_len_list entry in the TPA start\n"); - tpa_info->agg_state = QEDE_AGG_STATE_ERROR; + tpa_info->state = QEDE_AGG_STATE_ERROR; } } @@ -1197,7 +1312,7 @@ static void qede_gro_receive(struct qede_dev *edev, #ifdef CONFIG_INET if (skb_shinfo(skb)->gso_size) { - skb_set_network_header(skb, 0); + skb_reset_network_header(skb); switch (skb->protocol) { case htons(ETH_P_IP): @@ -1216,7 +1331,7 @@ static void qede_gro_receive(struct qede_dev *edev, send_skb: skb_record_rx_queue(skb, fp->rxq->rxq_id); - qede_skb_receive(edev, fp, skb, vlan_tag); + qede_skb_receive(edev, fp, fp->rxq, skb, vlan_tag); } static inline void qede_tpa_cont(struct qede_dev *edev, @@ -1253,7 +1368,7 @@ static void qede_tpa_end(struct qede_dev *edev, DP_ERR(edev, "Strange - TPA emd with more than a single len_list entry\n"); - if (unlikely(tpa_info->agg_state != QEDE_AGG_STATE_START)) + if (unlikely(tpa_info->state != QEDE_AGG_STATE_START)) goto err; /* Sanity */ @@ -1267,14 +1382,9 @@ static void qede_tpa_end(struct qede_dev *edev, le16_to_cpu(cqe->total_packet_len), skb->len); memcpy(skb->data, - page_address(tpa_info->start_buf.data) + - tpa_info->start_cqe.placement_offset + - tpa_info->start_buf.page_offset, - le16_to_cpu(tpa_info->start_cqe.len_on_first_bd)); - - /* Recycle [mapped] start buffer for the next replacement */ - tpa_info->replace_buf = tpa_info->start_buf; - tpa_info->replace_buf_mapping = tpa_info->start_buf_mapping; + page_address(tpa_info->buffer.data) + + tpa_info->start_cqe_placement_offset + + tpa_info->buffer.page_offset, tpa_info->start_cqe_bd_len); /* Finalize the SKB */ skb->protocol = eth_type_trans(skb, edev->ndev); @@ -1287,18 +1397,11 @@ static void qede_tpa_end(struct qede_dev *edev, qede_gro_receive(edev, fp, skb, tpa_info->vlan_tag); - tpa_info->agg_state = QEDE_AGG_STATE_NONE; + tpa_info->state = QEDE_AGG_STATE_NONE; return; err: - /* The BD starting the aggregation is still mapped; Re-use it for - * future aggregations [as replacement buffer] - */ - memcpy(&tpa_info->replace_buf, &tpa_info->start_buf, - sizeof(struct sw_rx_data)); - tpa_info->replace_buf_mapping = tpa_info->start_buf_mapping; - tpa_info->start_buf.data = NULL; - tpa_info->agg_state = QEDE_AGG_STATE_NONE; + tpa_info->state = QEDE_AGG_STATE_NONE; dev_kfree_skb_any(tpa_info->skb); tpa_info->skb = NULL; } @@ -1380,238 +1483,364 @@ static bool qede_pkt_is_ip_fragmented(struct eth_fast_path_rx_reg_cqe *cqe, return false; } -static int qede_rx_int(struct qede_fastpath *fp, int budget) +/* Return true iff packet is to be passed to stack */ +static bool qede_rx_xdp(struct qede_dev *edev, + struct qede_fastpath *fp, + struct qede_rx_queue *rxq, + struct bpf_prog *prog, + struct sw_rx_data *bd, + struct eth_fast_path_rx_reg_cqe *cqe) { - struct qede_dev *edev = fp->edev; - struct qede_rx_queue *rxq = fp->rxq; - - u16 hw_comp_cons, sw_comp_cons, sw_rx_index, parse_flag; - int rx_pkt = 0; - u8 csum_flag; + u16 len = le16_to_cpu(cqe->len_on_first_bd); + struct xdp_buff xdp; + enum xdp_action act; - hw_comp_cons = le16_to_cpu(*rxq->hw_cons_ptr); - sw_comp_cons = qed_chain_get_cons_idx(&rxq->rx_comp_ring); + xdp.data = page_address(bd->data) + cqe->placement_offset; + xdp.data_end = xdp.data + len; - /* Memory barrier to prevent the CPU from doing speculative reads of CQE - * / BD in the while-loop before reading hw_comp_cons. If the CQE is - * read before it is written by FW, then FW writes CQE and SB, and then - * the CPU reads the hw_comp_cons, it will use an old CQE. + /* Queues always have a full reset currently, so for the time + * being until there's atomic program replace just mark read + * side for map helpers. */ - rmb(); + rcu_read_lock(); + act = bpf_prog_run_xdp(prog, &xdp); + rcu_read_unlock(); - /* Loop to complete all indicated BDs */ - while (sw_comp_cons != hw_comp_cons) { - struct eth_fast_path_rx_reg_cqe *fp_cqe; - enum pkt_hash_types rxhash_type; - enum eth_rx_cqe_type cqe_type; - struct sw_rx_data *sw_rx_data; - union eth_rx_cqe *cqe; - struct sk_buff *skb; - struct page *data; - __le16 flags; - u16 len, pad; - u32 rx_hash; - - /* Get the CQE from the completion ring */ - cqe = (union eth_rx_cqe *) - qed_chain_consume(&rxq->rx_comp_ring); - cqe_type = cqe->fast_path_regular.type; - - if (unlikely(cqe_type == ETH_RX_CQE_TYPE_SLOW_PATH)) { - edev->ops->eth_cqe_completion( - edev->cdev, fp->id, - (struct eth_slow_path_rx_cqe *)cqe); - goto next_cqe; + if (act == XDP_PASS) + return true; + + /* Count number of packets not to be passed to stack */ + rxq->xdp_no_pass++; + + switch (act) { + case XDP_TX: + /* We need the replacement buffer before transmit. */ + if (qede_alloc_rx_buffer(rxq)) { + qede_recycle_rx_bd_ring(rxq, 1); + return false; } - if (cqe_type != ETH_RX_CQE_TYPE_REGULAR) { - switch (cqe_type) { - case ETH_RX_CQE_TYPE_TPA_START: - qede_tpa_start(edev, rxq, - &cqe->fast_path_tpa_start); - goto next_cqe; - case ETH_RX_CQE_TYPE_TPA_CONT: - qede_tpa_cont(edev, rxq, - &cqe->fast_path_tpa_cont); - goto next_cqe; - case ETH_RX_CQE_TYPE_TPA_END: - qede_tpa_end(edev, fp, - &cqe->fast_path_tpa_end); - goto next_rx_only; - default: - break; - } + /* Now if there's a transmission problem, we'd still have to + * throw current buffer, as replacement was already allocated. + */ + if (qede_xdp_xmit(edev, fp, bd, cqe->placement_offset, len)) { + dma_unmap_page(rxq->dev, bd->mapping, + PAGE_SIZE, DMA_BIDIRECTIONAL); + __free_page(bd->data); } - /* Get the data from the SW ring */ - sw_rx_index = rxq->sw_rx_cons & NUM_RX_BDS_MAX; - sw_rx_data = &rxq->sw_rx_ring[sw_rx_index]; - data = sw_rx_data->data; - - fp_cqe = &cqe->fast_path_regular; - len = le16_to_cpu(fp_cqe->len_on_first_bd); - pad = fp_cqe->placement_offset; - flags = cqe->fast_path_regular.pars_flags.flags; - - /* If this is an error packet then drop it */ - parse_flag = le16_to_cpu(flags); - - csum_flag = qede_check_csum(parse_flag); - if (unlikely(csum_flag == QEDE_CSUM_ERROR)) { - if (qede_pkt_is_ip_fragmented(&cqe->fast_path_regular, - parse_flag)) { - rxq->rx_ip_frags++; - goto alloc_skb; - } + /* Regardless, we've consumed an Rx BD */ + qede_rx_bd_ring_consume(rxq); + return false; - DP_NOTICE(edev, - "CQE in CONS = %u has error, flags = %x, dropping incoming packet\n", - sw_comp_cons, parse_flag); - rxq->rx_hw_errors++; - qede_recycle_rx_bd_ring(rxq, edev, fp_cqe->bd_num); - goto next_cqe; - } + default: + bpf_warn_invalid_xdp_action(act); + case XDP_ABORTED: + case XDP_DROP: + qede_recycle_rx_bd_ring(rxq, cqe->bd_num); + } -alloc_skb: - skb = netdev_alloc_skb(edev->ndev, QEDE_RX_HDR_SIZE); - if (unlikely(!skb)) { - DP_NOTICE(edev, - "skb allocation failed, dropping incoming packet\n"); - qede_recycle_rx_bd_ring(rxq, edev, fp_cqe->bd_num); - rxq->rx_alloc_errors++; - goto next_cqe; + return false; +} + +static struct sk_buff *qede_rx_allocate_skb(struct qede_dev *edev, + struct qede_rx_queue *rxq, + struct sw_rx_data *bd, u16 len, + u16 pad) +{ + unsigned int offset = bd->page_offset; + struct skb_frag_struct *frag; + struct page *page = bd->data; + unsigned int pull_len; + struct sk_buff *skb; + unsigned char *va; + + /* Allocate a new SKB with a sufficient large header len */ + skb = netdev_alloc_skb(edev->ndev, QEDE_RX_HDR_SIZE); + if (unlikely(!skb)) + return NULL; + + /* Copy data into SKB - if it's small, we can simply copy it and + * re-use the already allcoated & mapped memory. + */ + if (len + pad <= edev->rx_copybreak) { + memcpy(skb_put(skb, len), + page_address(page) + pad + offset, len); + qede_reuse_page(rxq, bd); + goto out; + } + + frag = &skb_shinfo(skb)->frags[0]; + + skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, + page, pad + offset, len, rxq->rx_buf_seg_size); + + va = skb_frag_address(frag); + pull_len = eth_get_headlen(va, QEDE_RX_HDR_SIZE); + + /* Align the pull_len to optimize memcpy */ + memcpy(skb->data, va, ALIGN(pull_len, sizeof(long))); + + /* Correct the skb & frag sizes offset after the pull */ + skb_frag_size_sub(frag, pull_len); + frag->page_offset += pull_len; + skb->data_len -= pull_len; + skb->tail += pull_len; + + if (unlikely(qede_realloc_rx_buffer(rxq, bd))) { + /* Incr page ref count to reuse on allocation failure so + * that it doesn't get freed while freeing SKB [as its + * already mapped there]. + */ + page_ref_inc(page); + dev_kfree_skb_any(skb); + return NULL; + } + +out: + /* We've consumed the first BD and prepared an SKB */ + qede_rx_bd_ring_consume(rxq); + return skb; +} + +static int qede_rx_build_jumbo(struct qede_dev *edev, + struct qede_rx_queue *rxq, + struct sk_buff *skb, + struct eth_fast_path_rx_reg_cqe *cqe, + u16 first_bd_len) +{ + u16 pkt_len = le16_to_cpu(cqe->pkt_len); + struct sw_rx_data *bd; + u16 bd_cons_idx; + u8 num_frags; + + pkt_len -= first_bd_len; + + /* We've already used one BD for the SKB. Now take care of the rest */ + for (num_frags = cqe->bd_num - 1; num_frags > 0; num_frags--) { + u16 cur_size = pkt_len > rxq->rx_buf_size ? rxq->rx_buf_size : + pkt_len; + + if (unlikely(!cur_size)) { + DP_ERR(edev, + "Still got %d BDs for mapping jumbo, but length became 0\n", + num_frags); + goto out; } - /* Copy data into SKB */ - if (len + pad <= edev->rx_copybreak) { - memcpy(skb_put(skb, len), - page_address(data) + pad + - sw_rx_data->page_offset, len); - qede_reuse_page(edev, rxq, sw_rx_data); + /* We need a replacement buffer for each BD */ + if (unlikely(qede_alloc_rx_buffer(rxq))) + goto out; + + /* Now that we've allocated the replacement buffer, + * we can safely consume the next BD and map it to the SKB. + */ + bd_cons_idx = rxq->sw_rx_cons & NUM_RX_BDS_MAX; + bd = &rxq->sw_rx_ring[bd_cons_idx]; + qede_rx_bd_ring_consume(rxq); + + dma_unmap_page(rxq->dev, bd->mapping, + PAGE_SIZE, DMA_FROM_DEVICE); + + skb_fill_page_desc(skb, skb_shinfo(skb)->nr_frags++, + bd->data, 0, cur_size); + + skb->truesize += PAGE_SIZE; + skb->data_len += cur_size; + skb->len += cur_size; + pkt_len -= cur_size; + } + + if (unlikely(pkt_len)) + DP_ERR(edev, + "Mapped all BDs of jumbo, but still have %d bytes\n", + pkt_len); + +out: + return num_frags; +} + +static int qede_rx_process_tpa_cqe(struct qede_dev *edev, + struct qede_fastpath *fp, + struct qede_rx_queue *rxq, + union eth_rx_cqe *cqe, + enum eth_rx_cqe_type type) +{ + switch (type) { + case ETH_RX_CQE_TYPE_TPA_START: + qede_tpa_start(edev, rxq, &cqe->fast_path_tpa_start); + return 0; + case ETH_RX_CQE_TYPE_TPA_CONT: + qede_tpa_cont(edev, rxq, &cqe->fast_path_tpa_cont); + return 0; + case ETH_RX_CQE_TYPE_TPA_END: + qede_tpa_end(edev, fp, &cqe->fast_path_tpa_end); + return 1; + default: + return 0; + } +} + +static int qede_rx_process_cqe(struct qede_dev *edev, + struct qede_fastpath *fp, + struct qede_rx_queue *rxq) +{ + struct bpf_prog *xdp_prog = READ_ONCE(rxq->xdp_prog); + struct eth_fast_path_rx_reg_cqe *fp_cqe; + u16 len, pad, bd_cons_idx, parse_flag; + enum eth_rx_cqe_type cqe_type; + union eth_rx_cqe *cqe; + struct sw_rx_data *bd; + struct sk_buff *skb; + __le16 flags; + u8 csum_flag; + + /* Get the CQE from the completion ring */ + cqe = (union eth_rx_cqe *)qed_chain_consume(&rxq->rx_comp_ring); + cqe_type = cqe->fast_path_regular.type; + + /* Process an unlikely slowpath event */ + if (unlikely(cqe_type == ETH_RX_CQE_TYPE_SLOW_PATH)) { + struct eth_slow_path_rx_cqe *sp_cqe; + + sp_cqe = (struct eth_slow_path_rx_cqe *)cqe; + edev->ops->eth_cqe_completion(edev->cdev, fp->id, sp_cqe); + return 0; + } + + /* Handle TPA cqes */ + if (cqe_type != ETH_RX_CQE_TYPE_REGULAR) + return qede_rx_process_tpa_cqe(edev, fp, rxq, cqe, cqe_type); + + /* Get the data from the SW ring; Consume it only after it's evident + * we wouldn't recycle it. + */ + bd_cons_idx = rxq->sw_rx_cons & NUM_RX_BDS_MAX; + bd = &rxq->sw_rx_ring[bd_cons_idx]; + + fp_cqe = &cqe->fast_path_regular; + len = le16_to_cpu(fp_cqe->len_on_first_bd); + pad = fp_cqe->placement_offset; + + /* Run eBPF program if one is attached */ + if (xdp_prog) + if (!qede_rx_xdp(edev, fp, rxq, xdp_prog, bd, fp_cqe)) + return 1; + + /* If this is an error packet then drop it */ + flags = cqe->fast_path_regular.pars_flags.flags; + parse_flag = le16_to_cpu(flags); + + csum_flag = qede_check_csum(parse_flag); + if (unlikely(csum_flag == QEDE_CSUM_ERROR)) { + if (qede_pkt_is_ip_fragmented(fp_cqe, parse_flag)) { + rxq->rx_ip_frags++; } else { - struct skb_frag_struct *frag; - unsigned int pull_len; - unsigned char *va; - - frag = &skb_shinfo(skb)->frags[0]; - - skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, data, - pad + sw_rx_data->page_offset, - len, rxq->rx_buf_seg_size); - - va = skb_frag_address(frag); - pull_len = eth_get_headlen(va, QEDE_RX_HDR_SIZE); - - /* Align the pull_len to optimize memcpy */ - memcpy(skb->data, va, ALIGN(pull_len, sizeof(long))); - - skb_frag_size_sub(frag, pull_len); - frag->page_offset += pull_len; - skb->data_len -= pull_len; - skb->tail += pull_len; - - if (unlikely(qede_realloc_rx_buffer(edev, rxq, - sw_rx_data))) { - DP_ERR(edev, "Failed to allocate rx buffer\n"); - /* Incr page ref count to reuse on allocation - * failure so that it doesn't get freed while - * freeing SKB. - */ - - page_ref_inc(sw_rx_data->data); - rxq->rx_alloc_errors++; - qede_recycle_rx_bd_ring(rxq, edev, - fp_cqe->bd_num); - dev_kfree_skb_any(skb); - goto next_cqe; - } + DP_NOTICE(edev, + "CQE has error, flags = %x, dropping incoming packet\n", + parse_flag); + rxq->rx_hw_errors++; + qede_recycle_rx_bd_ring(rxq, fp_cqe->bd_num); + return 0; } + } - qede_rx_bd_ring_consume(rxq); + /* Basic validation passed; Need to prepare an SKB. This would also + * guarantee to finally consume the first BD upon success. + */ + skb = qede_rx_allocate_skb(edev, rxq, bd, len, pad); + if (!skb) { + rxq->rx_alloc_errors++; + qede_recycle_rx_bd_ring(rxq, fp_cqe->bd_num); + return 0; + } - if (fp_cqe->bd_num != 1) { - u16 pkt_len = le16_to_cpu(fp_cqe->pkt_len); - u8 num_frags; - - pkt_len -= len; - - for (num_frags = fp_cqe->bd_num - 1; num_frags > 0; - num_frags--) { - u16 cur_size = pkt_len > rxq->rx_buf_size ? - rxq->rx_buf_size : pkt_len; - if (unlikely(!cur_size)) { - DP_ERR(edev, - "Still got %d BDs for mapping jumbo, but length became 0\n", - num_frags); - qede_recycle_rx_bd_ring(rxq, edev, - num_frags); - dev_kfree_skb_any(skb); - goto next_cqe; - } - - if (unlikely(qede_alloc_rx_buffer(edev, rxq))) { - qede_recycle_rx_bd_ring(rxq, edev, - num_frags); - dev_kfree_skb_any(skb); - goto next_cqe; - } - - sw_rx_index = rxq->sw_rx_cons & NUM_RX_BDS_MAX; - sw_rx_data = &rxq->sw_rx_ring[sw_rx_index]; - qede_rx_bd_ring_consume(rxq); - - dma_unmap_page(&edev->pdev->dev, - sw_rx_data->mapping, - PAGE_SIZE, DMA_FROM_DEVICE); - - skb_fill_page_desc(skb, - skb_shinfo(skb)->nr_frags++, - sw_rx_data->data, 0, - cur_size); - - skb->truesize += PAGE_SIZE; - skb->data_len += cur_size; - skb->len += cur_size; - pkt_len -= cur_size; - } + /* In case of Jumbo packet, several PAGE_SIZEd buffers will be pointed + * by a single cqe. + */ + if (fp_cqe->bd_num > 1) { + u16 unmapped_frags = qede_rx_build_jumbo(edev, rxq, skb, + fp_cqe, len); - if (unlikely(pkt_len)) - DP_ERR(edev, - "Mapped all BDs of jumbo, but still have %d bytes\n", - pkt_len); + if (unlikely(unmapped_frags > 0)) { + qede_recycle_rx_bd_ring(rxq, unmapped_frags); + dev_kfree_skb_any(skb); + return 0; } + } - skb->protocol = eth_type_trans(skb, edev->ndev); + /* The SKB contains all the data. Now prepare meta-magic */ + skb->protocol = eth_type_trans(skb, edev->ndev); + qede_get_rxhash(skb, fp_cqe->bitfields, fp_cqe->rss_hash); + qede_set_skb_csum(skb, csum_flag); + skb_record_rx_queue(skb, rxq->rxq_id); - rx_hash = qede_get_rxhash(edev, fp_cqe->bitfields, - fp_cqe->rss_hash, &rxhash_type); + /* SKB is prepared - pass it to stack */ + qede_skb_receive(edev, fp, rxq, skb, le16_to_cpu(fp_cqe->vlan_tag)); - skb_set_hash(skb, rx_hash, rxhash_type); + return 1; +} - qede_set_skb_csum(skb, csum_flag); +static int qede_rx_int(struct qede_fastpath *fp, int budget) +{ + struct qede_rx_queue *rxq = fp->rxq; + struct qede_dev *edev = fp->edev; + u16 hw_comp_cons, sw_comp_cons; + int work_done = 0; - skb_record_rx_queue(skb, fp->rxq->rxq_id); + hw_comp_cons = le16_to_cpu(*rxq->hw_cons_ptr); + sw_comp_cons = qed_chain_get_cons_idx(&rxq->rx_comp_ring); - qede_skb_receive(edev, fp, skb, le16_to_cpu(fp_cqe->vlan_tag)); -next_rx_only: - rx_pkt++; + /* Memory barrier to prevent the CPU from doing speculative reads of CQE + * / BD in the while-loop before reading hw_comp_cons. If the CQE is + * read before it is written by FW, then FW writes CQE and SB, and then + * the CPU reads the hw_comp_cons, it will use an old CQE. + */ + rmb(); -next_cqe: /* don't consume bd rx buffer */ + /* Loop to complete all indicated BDs */ + while ((sw_comp_cons != hw_comp_cons) && (work_done < budget)) { + qede_rx_process_cqe(edev, fp, rxq); qed_chain_recycle_consumed(&rxq->rx_comp_ring); sw_comp_cons = qed_chain_get_cons_idx(&rxq->rx_comp_ring); - /* CR TPA - revisit how to handle budget in TPA perhaps - * increase on "end" - */ - if (rx_pkt == budget) - break; - } /* repeat while sw_comp_cons != hw_comp_cons... */ + work_done++; + } /* Update producers */ qede_update_rx_prod(edev, rxq); - rxq->rcv_pkts += rx_pkt; + return work_done; +} + +static bool qede_poll_is_more_work(struct qede_fastpath *fp) +{ + qed_sb_update_sb_idx(fp->sb_info); + + /* *_has_*_work() reads the status block, thus we need to ensure that + * status block indices have been actually read (qed_sb_update_sb_idx) + * prior to this check (*_has_*_work) so that we won't write the + * "newer" value of the status block to HW (if there was a DMA right + * after qede_has_rx_work and if there is no rmb, the memory reading + * (qed_sb_update_sb_idx) may be postponed to right before *_ack_sb). + * In this case there will never be another interrupt until there is + * another update of the status block, while there is still unhandled + * work. + */ + rmb(); + + if (likely(fp->type & QEDE_FASTPATH_RX)) + if (qede_has_rx_work(fp->rxq)) + return true; - return rx_pkt; + if (fp->type & QEDE_FASTPATH_XDP) + if (qede_txq_has_work(fp->xdp_tx)) + return true; + + if (likely(fp->type & QEDE_FASTPATH_TX)) + if (qede_txq_has_work(fp->txq)) + return true; + + return false; } static int qede_poll(struct napi_struct *napi, int budget) @@ -1620,48 +1849,35 @@ static int qede_poll(struct napi_struct *napi, int budget) napi); struct qede_dev *edev = fp->edev; int rx_work_done = 0; - u8 tc; - for (tc = 0; tc < edev->num_tc; tc++) - if (likely(fp->type & QEDE_FASTPATH_TX) && - qede_txq_has_work(&fp->txqs[tc])) - qede_tx_int(edev, &fp->txqs[tc]); + if (likely(fp->type & QEDE_FASTPATH_TX) && qede_txq_has_work(fp->txq)) + qede_tx_int(edev, fp->txq); + + if ((fp->type & QEDE_FASTPATH_XDP) && qede_txq_has_work(fp->xdp_tx)) + qede_xdp_tx_int(edev, fp->xdp_tx); rx_work_done = (likely(fp->type & QEDE_FASTPATH_RX) && qede_has_rx_work(fp->rxq)) ? qede_rx_int(fp, budget) : 0; if (rx_work_done < budget) { - qed_sb_update_sb_idx(fp->sb_info); - /* *_has_*_work() reads the status block, - * thus we need to ensure that status block indices - * have been actually read (qed_sb_update_sb_idx) - * prior to this check (*_has_*_work) so that - * we won't write the "newer" value of the status block - * to HW (if there was a DMA right after - * qede_has_rx_work and if there is no rmb, the memory - * reading (qed_sb_update_sb_idx) may be postponed - * to right before *_ack_sb). In this case there - * will never be another interrupt until there is - * another update of the status block, while there - * is still unhandled work. - */ - rmb(); - - /* Fall out from the NAPI loop if needed */ - if (!((likely(fp->type & QEDE_FASTPATH_RX) && - qede_has_rx_work(fp->rxq)) || - (likely(fp->type & QEDE_FASTPATH_TX) && - qede_has_tx_work(fp)))) { + if (!qede_poll_is_more_work(fp)) { napi_complete(napi); /* Update and reenable interrupts */ - qed_sb_ack(fp->sb_info, IGU_INT_ENABLE, - 1 /*update*/); + qed_sb_ack(fp->sb_info, IGU_INT_ENABLE, 1); } else { rx_work_done = budget; } } + if (fp->xdp_xmit) { + u16 xdp_prod = qed_chain_get_prod_idx(&fp->xdp_tx->tx_pbl); + + fp->xdp_xmit = 0; + fp->xdp_tx->tx_db.data.bd_prod = cpu_to_le16(xdp_prod); + qede_update_tx_producer(fp->xdp_tx); + } + return rx_work_done; } @@ -1912,7 +2128,7 @@ static int qede_vlan_rx_add_vid(struct net_device *dev, __be16 proto, u16 vid) { struct qede_dev *edev = netdev_priv(dev); struct qede_vlan *vlan, *tmp; - int rc; + int rc = 0; DP_VERBOSE(edev, NETIF_MSG_IFUP, "Adding vlan 0x%04x\n", vid); @@ -1936,6 +2152,7 @@ static int qede_vlan_rx_add_vid(struct net_device *dev, __be16 proto, u16 vid) } /* If interface is down, cache this VLAN ID and return */ + __qede_lock(edev); if (edev->state != QEDE_STATE_OPEN) { DP_VERBOSE(edev, NETIF_MSG_IFDOWN, "Interface is down, VLAN %d will be configured when interface is up\n", @@ -1943,8 +2160,7 @@ static int qede_vlan_rx_add_vid(struct net_device *dev, __be16 proto, u16 vid) if (vid != 0) edev->non_configured_vlans++; list_add(&vlan->list, &edev->vlan_list); - - return 0; + goto out; } /* Check for the filter limit. @@ -1960,7 +2176,7 @@ static int qede_vlan_rx_add_vid(struct net_device *dev, __be16 proto, u16 vid) DP_ERR(edev, "Failed to configure VLAN %d\n", vlan->vid); kfree(vlan); - return -EINVAL; + goto out; } vlan->configured = true; @@ -1977,7 +2193,9 @@ static int qede_vlan_rx_add_vid(struct net_device *dev, __be16 proto, u16 vid) list_add(&vlan->list, &edev->vlan_list); - return 0; +out: + __qede_unlock(edev); + return rc; } static void qede_del_vlan_from_list(struct qede_dev *edev, @@ -2054,11 +2272,12 @@ static int qede_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid) { struct qede_dev *edev = netdev_priv(dev); struct qede_vlan *vlan = NULL; - int rc; + int rc = 0; DP_VERBOSE(edev, NETIF_MSG_IFDOWN, "Removing vlan 0x%04x\n", vid); /* Find whether entry exists */ + __qede_lock(edev); list_for_each_entry(vlan, &edev->vlan_list, list) if (vlan->vid == vid) break; @@ -2066,7 +2285,7 @@ static int qede_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid) if (!vlan || (vlan->vid != vid)) { DP_VERBOSE(edev, (NETIF_MSG_IFUP | NETIF_MSG_IFDOWN), "Vlan isn't configured\n"); - return 0; + goto out; } if (edev->state != QEDE_STATE_OPEN) { @@ -2076,7 +2295,7 @@ static int qede_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid) DP_VERBOSE(edev, NETIF_MSG_IFDOWN, "Interface is down, removing VLAN from list only\n"); qede_del_vlan_from_list(edev, vlan); - return 0; + goto out; } /* Remove vlan */ @@ -2085,7 +2304,7 @@ static int qede_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid) vid); if (rc) { DP_ERR(edev, "Failed to remove VLAN %d\n", vid); - return -EINVAL; + goto out; } } @@ -2096,6 +2315,8 @@ static int qede_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid) */ rc = qede_configure_vlan_filters(edev); +out: + __qede_unlock(edev); return rc; } @@ -2125,7 +2346,13 @@ static void qede_vlan_mark_nonconfigured(struct qede_dev *edev) edev->accept_any_vlan = false; } -static int qede_set_features(struct net_device *dev, netdev_features_t features) +static void qede_set_features_reload(struct qede_dev *edev, + struct qede_reload_args *args) +{ + edev->ndev->features = args->u.features; +} + +int qede_set_features(struct net_device *dev, netdev_features_t features) { struct qede_dev *edev = netdev_priv(dev); netdev_features_t changes = features ^ dev->features; @@ -2139,9 +2366,23 @@ static int qede_set_features(struct net_device *dev, netdev_features_t features) need_reload = edev->gro_disable; } - if (need_reload && netif_running(edev->ndev)) { - dev->features = features; - qede_reload(edev, NULL, NULL); + if (need_reload) { + struct qede_reload_args args; + + args.u.features = features; + args.func = &qede_set_features_reload; + + /* Make sure that we definitely need to reload. + * In case of an eBPF attached program, there will be no FW + * aggregations, so no need to actually reload. + */ + __qede_lock(edev); + if (edev->xdp_prog) + args.func(edev, &args); + else + qede_reload(edev, &args, true); + __qede_unlock(edev); + return 1; } @@ -2218,6 +2459,82 @@ static void qede_udp_tunnel_del(struct net_device *dev, schedule_delayed_work(&edev->sp_task, 0); } +/* 8B udp header + 8B base tunnel header + 32B option length */ +#define QEDE_MAX_TUN_HDR_LEN 48 + +static netdev_features_t qede_features_check(struct sk_buff *skb, + struct net_device *dev, + netdev_features_t features) +{ + if (skb->encapsulation) { + u8 l4_proto = 0; + + switch (vlan_get_protocol(skb)) { + case htons(ETH_P_IP): + l4_proto = ip_hdr(skb)->protocol; + break; + case htons(ETH_P_IPV6): + l4_proto = ipv6_hdr(skb)->nexthdr; + break; + default: + return features; + } + + /* Disable offloads for geneve tunnels, as HW can't parse + * the geneve header which has option length greater than 32B. + */ + if ((l4_proto == IPPROTO_UDP) && + ((skb_inner_mac_header(skb) - + skb_transport_header(skb)) > QEDE_MAX_TUN_HDR_LEN)) + return features & ~(NETIF_F_CSUM_MASK | + NETIF_F_GSO_MASK); + } + + return features; +} + +static void qede_xdp_reload_func(struct qede_dev *edev, + struct qede_reload_args *args) +{ + struct bpf_prog *old; + + old = xchg(&edev->xdp_prog, args->u.new_prog); + if (old) + bpf_prog_put(old); +} + +static int qede_xdp_set(struct qede_dev *edev, struct bpf_prog *prog) +{ + struct qede_reload_args args; + + if (prog && prog->xdp_adjust_head) { + DP_ERR(edev, "Does not support bpf_xdp_adjust_head()\n"); + return -EOPNOTSUPP; + } + + /* If we're called, there was already a bpf reference increment */ + args.func = &qede_xdp_reload_func; + args.u.new_prog = prog; + qede_reload(edev, &args, false); + + return 0; +} + +static int qede_xdp(struct net_device *dev, struct netdev_xdp *xdp) +{ + struct qede_dev *edev = netdev_priv(dev); + + switch (xdp->command) { + case XDP_SETUP_PROG: + return qede_xdp_set(edev, xdp->prog); + case XDP_QUERY_PROG: + xdp->prog_attached = !!edev->xdp_prog; + return 0; + default: + return -EINVAL; + } +} + static const struct net_device_ops qede_netdev_ops = { .ndo_open = qede_open, .ndo_stop = qede_close, @@ -2242,6 +2559,8 @@ static const struct net_device_ops qede_netdev_ops = { #endif .ndo_udp_tunnel_add = qede_udp_tunnel_add, .ndo_udp_tunnel_del = qede_udp_tunnel_del, + .ndo_features_check = qede_features_check, + .ndo_xdp = qede_xdp, }; /* ------------------------------------------------------------------------- @@ -2282,8 +2601,6 @@ static struct qede_dev *qede_alloc_etherdev(struct qed_dev *cdev, memset(&edev->stats, 0, sizeof(edev->stats)); memcpy(&edev->dev_info, info, sizeof(*info)); - edev->num_tc = edev->dev_info.num_tc; - INIT_LIST_HEAD(&edev->vlan_list); return edev; @@ -2308,6 +2625,8 @@ static void qede_init_ndev(struct qede_dev *edev) qede_set_ethtool_ops(ndev); + ndev->priv_flags |= IFF_UNICAST_FLT; + /* user-changeble features */ hw_features = NETIF_F_GRO | NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | @@ -2315,11 +2634,14 @@ static void qede_init_ndev(struct qede_dev *edev) /* Encap features*/ hw_features |= NETIF_F_GSO_GRE | NETIF_F_GSO_UDP_TUNNEL | - NETIF_F_TSO_ECN; + NETIF_F_TSO_ECN | NETIF_F_GSO_UDP_TUNNEL_CSUM | + NETIF_F_GSO_GRE_CSUM; ndev->hw_enc_features = NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_SG | NETIF_F_TSO | NETIF_F_TSO_ECN | NETIF_F_TSO6 | NETIF_F_GSO_GRE | - NETIF_F_GSO_UDP_TUNNEL | NETIF_F_RXCSUM; + NETIF_F_GSO_UDP_TUNNEL | NETIF_F_RXCSUM | + NETIF_F_GSO_UDP_TUNNEL_CSUM | + NETIF_F_GSO_GRE_CSUM; ndev->vlan_features = hw_features | NETIF_F_RXHASH | NETIF_F_RXCSUM | NETIF_F_HIGHDMA; @@ -2329,8 +2651,14 @@ static void qede_init_ndev(struct qede_dev *edev) ndev->hw_features = hw_features; + /* MTU range: 46 - 9600 */ + ndev->min_mtu = ETH_ZLEN - ETH_HLEN; + ndev->max_mtu = QEDE_MAX_JUMBO_PACKET_SIZE; + /* Set network device HW mac */ ether_addr_copy(edev->ndev->dev_addr, edev->dev_info.common.hw_mac); + + ndev->mtu = edev->dev_info.common.mtu; } /* This function converts from 32b param to two params of level and module @@ -2370,7 +2698,8 @@ static void qede_free_fp_array(struct qede_dev *edev) kfree(fp->sb_info); kfree(fp->rxq); - kfree(fp->txqs); + kfree(fp->xdp_tx); + kfree(fp->txq); } kfree(edev->fp_array); } @@ -2403,7 +2732,7 @@ static int qede_alloc_fp_array(struct qede_dev *edev) for_each_queue(i) { fp = &edev->fp_array[i]; - fp->sb_info = kcalloc(1, sizeof(*fp->sb_info), GFP_KERNEL); + fp->sb_info = kzalloc(sizeof(*fp->sb_info), GFP_KERNEL); if (!fp->sb_info) { DP_NOTICE(edev, "sb info struct allocation failed\n"); goto err; @@ -2420,21 +2749,22 @@ static int qede_alloc_fp_array(struct qede_dev *edev) } if (fp->type & QEDE_FASTPATH_TX) { - fp->txqs = kcalloc(edev->num_tc, sizeof(*fp->txqs), - GFP_KERNEL); - if (!fp->txqs) { - DP_NOTICE(edev, - "TXQ array allocation failed\n"); + fp->txq = kzalloc(sizeof(*fp->txq), GFP_KERNEL); + if (!fp->txq) goto err; - } } if (fp->type & QEDE_FASTPATH_RX) { - fp->rxq = kcalloc(1, sizeof(*fp->rxq), GFP_KERNEL); - if (!fp->rxq) { - DP_NOTICE(edev, - "RXQ struct allocation failed\n"); + fp->rxq = kzalloc(sizeof(*fp->rxq), GFP_KERNEL); + if (!fp->rxq) goto err; + + if (edev->xdp_prog) { + fp->xdp_tx = kzalloc(sizeof(*fp->xdp_tx), + GFP_KERNEL); + if (!fp->xdp_tx) + goto err; + fp->type |= QEDE_FASTPATH_XDP; } } } @@ -2451,12 +2781,11 @@ static void qede_sp_task(struct work_struct *work) sp_task.work); struct qed_dev *cdev = edev->cdev; - mutex_lock(&edev->qede_lock); + __qede_lock(edev); - if (edev->state == QEDE_STATE_OPEN) { - if (test_and_clear_bit(QEDE_SP_RX_MODE, &edev->sp_flags)) + if (test_and_clear_bit(QEDE_SP_RX_MODE, &edev->sp_flags)) + if (edev->state == QEDE_STATE_OPEN) qede_config_rx_mode(edev->ndev); - } if (test_and_clear_bit(QEDE_SP_VXLAN_PORT_CONFIG, &edev->sp_flags)) { struct qed_tunn_params tunn_params; @@ -2476,16 +2805,16 @@ static void qede_sp_task(struct work_struct *work) qed_ops->tunn_config(cdev, &tunn_params); } - mutex_unlock(&edev->qede_lock); + __qede_unlock(edev); } static void qede_update_pf_params(struct qed_dev *cdev) { struct qed_pf_params pf_params; - /* 64 rx + 64 tx */ + /* 64 rx + 64 tx + 64 XDP */ memset(&pf_params, 0, sizeof(struct qed_pf_params)); - pf_params.eth_pf_params.num_cons = 128; + pf_params.eth_pf_params.num_cons = 192; qed_ops->common->update_pf_params(cdev, &pf_params); } @@ -2634,10 +2963,16 @@ static void __qede_remove(struct pci_dev *pdev, enum qede_remove_mode mode) pci_set_drvdata(pdev, NULL); + /* Release edev's reference to XDP's bpf if such exist */ + if (edev->xdp_prog) + bpf_prog_put(edev->xdp_prog); + free_netdev(ndev); /* Use global ops since we've freed edev */ qed_ops->common->slowpath_stop(cdev); + if (system_state == SYSTEM_POWER_OFF) + return; qed_ops->common->remove(cdev); dev_info(&pdev->dev, "Ending qede_remove successfully\n"); @@ -2648,6 +2983,11 @@ static void qede_remove(struct pci_dev *pdev) __qede_remove(pdev, QEDE_REMOVE_NORMAL); } +static void qede_shutdown(struct pci_dev *pdev) +{ + __qede_remove(pdev, QEDE_REMOVE_NORMAL); +} + /* ------------------------------------------------------------------------- * START OF LOAD / UNLOAD * ------------------------------------------------------------------------- @@ -2731,7 +3071,7 @@ static void qede_free_rx_buffers(struct qede_dev *edev, data = rx_buf->data; dma_unmap_page(&edev->pdev->dev, - rx_buf->mapping, PAGE_SIZE, DMA_FROM_DEVICE); + rx_buf->mapping, PAGE_SIZE, rxq->data_direction); rx_buf->data = NULL; __free_page(data); @@ -2747,7 +3087,7 @@ static void qede_free_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq) for (i = 0; i < ETH_TPA_MAX_AGGS_NUM; i++) { struct qede_agg_info *tpa_info = &rxq->tpa_info[i]; - struct sw_rx_data *replace_buf = &tpa_info->replace_buf; + struct sw_rx_data *replace_buf = &tpa_info->buffer; if (replace_buf->data) { dma_unmap_page(&edev->pdev->dev, @@ -2773,52 +3113,15 @@ static void qede_free_mem_rxq(struct qede_dev *edev, struct qede_rx_queue *rxq) edev->ops->common->chain_free(edev->cdev, &rxq->rx_comp_ring); } -static int qede_alloc_rx_buffer(struct qede_dev *edev, - struct qede_rx_queue *rxq) -{ - struct sw_rx_data *sw_rx_data; - struct eth_rx_bd *rx_bd; - dma_addr_t mapping; - struct page *data; - - data = alloc_pages(GFP_ATOMIC, 0); - if (unlikely(!data)) { - DP_NOTICE(edev, "Failed to allocate Rx data [page]\n"); - return -ENOMEM; - } - - /* Map the entire page as it would be used - * for multiple RX buffer segment size mapping. - */ - mapping = dma_map_page(&edev->pdev->dev, data, 0, - PAGE_SIZE, DMA_FROM_DEVICE); - if (unlikely(dma_mapping_error(&edev->pdev->dev, mapping))) { - __free_page(data); - DP_NOTICE(edev, "Failed to map Rx buffer\n"); - return -ENOMEM; - } - - sw_rx_data = &rxq->sw_rx_ring[rxq->sw_rx_prod & NUM_RX_BDS_MAX]; - sw_rx_data->page_offset = 0; - sw_rx_data->data = data; - sw_rx_data->mapping = mapping; - - /* Advance PROD and get BD pointer */ - rx_bd = (struct eth_rx_bd *)qed_chain_produce(&rxq->rx_bd_ring); - WARN_ON(!rx_bd); - rx_bd->addr.hi = cpu_to_le32(upper_32_bits(mapping)); - rx_bd->addr.lo = cpu_to_le32(lower_32_bits(mapping)); - - rxq->sw_rx_prod++; - - return 0; -} - static int qede_alloc_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq) { dma_addr_t mapping; int i; + /* Don't perform FW aggregations in case of XDP */ + if (edev->xdp_prog) + edev->gro_disable = 1; + if (edev->gro_disable) return 0; @@ -2829,7 +3132,7 @@ static int qede_alloc_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq) for (i = 0; i < ETH_TPA_MAX_AGGS_NUM; i++) { struct qede_agg_info *tpa_info = &rxq->tpa_info[i]; - struct sw_rx_data *replace_buf = &tpa_info->replace_buf; + struct sw_rx_data *replace_buf = &tpa_info->buffer; replace_buf->data = alloc_pages(GFP_ATOMIC, 0); if (unlikely(!replace_buf->data)) { @@ -2839,7 +3142,7 @@ static int qede_alloc_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq) } mapping = dma_map_page(&edev->pdev->dev, replace_buf->data, 0, - rxq->rx_buf_size, DMA_FROM_DEVICE); + PAGE_SIZE, DMA_FROM_DEVICE); if (unlikely(dma_mapping_error(&edev->pdev->dev, mapping))) { DP_NOTICE(edev, "Failed to map TPA replacement buffer\n"); @@ -2847,10 +3150,9 @@ static int qede_alloc_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq) } replace_buf->mapping = mapping; - tpa_info->replace_buf.page_offset = 0; - - tpa_info->replace_buf_mapping = mapping; - tpa_info->agg_state = QEDE_AGG_STATE_NONE; + tpa_info->buffer.page_offset = 0; + tpa_info->buffer_mapping = mapping; + tpa_info->state = QEDE_AGG_STATE_NONE; } return 0; @@ -2872,8 +3174,13 @@ static int qede_alloc_mem_rxq(struct qede_dev *edev, struct qede_rx_queue *rxq) if (rxq->rx_buf_size > PAGE_SIZE) rxq->rx_buf_size = PAGE_SIZE; - /* Segment size to spilt a page in multiple equal parts */ - rxq->rx_buf_seg_size = roundup_pow_of_two(rxq->rx_buf_size); + /* Segment size to spilt a page in multiple equal parts, + * unless XDP is used in which case we'd use the entire page. + */ + if (!edev->xdp_prog) + rxq->rx_buf_seg_size = roundup_pow_of_two(rxq->rx_buf_size); + else + rxq->rx_buf_seg_size = PAGE_SIZE; /* Allocate the parallel driver ring for Rx buffers */ size = sizeof(*rxq->sw_rx_ring) * RX_RING_SIZE; @@ -2909,7 +3216,7 @@ static int qede_alloc_mem_rxq(struct qede_dev *edev, struct qede_rx_queue *rxq) /* Allocate buffers for the Rx ring */ for (i = 0; i < rxq->num_rx_buffers; i++) { - rc = qede_alloc_rx_buffer(edev, rxq); + rc = qede_alloc_rx_buffer(rxq); if (rc) { DP_ERR(edev, "Rx buffers allocation failed at index %d\n", i); @@ -2925,7 +3232,10 @@ err: static void qede_free_mem_txq(struct qede_dev *edev, struct qede_tx_queue *txq) { /* Free the parallel SW ring */ - kfree(txq->sw_tx_ring); + if (txq->is_xdp) + kfree(txq->sw_tx_ring.pages); + else + kfree(txq->sw_tx_ring.skbs); /* Free the real RQ ring used by FW */ edev->ops->common->chain_free(edev->cdev, &txq->tx_pbl); @@ -2934,17 +3244,22 @@ static void qede_free_mem_txq(struct qede_dev *edev, struct qede_tx_queue *txq) /* This function allocates all memory needed per Tx queue */ static int qede_alloc_mem_txq(struct qede_dev *edev, struct qede_tx_queue *txq) { - int size, rc; union eth_tx_bd_types *p_virt; + int size, rc; txq->num_tx_buffers = edev->q_num_tx_buffers; /* Allocate the parallel driver ring for Tx buffers */ - size = sizeof(*txq->sw_tx_ring) * TX_RING_SIZE; - txq->sw_tx_ring = kzalloc(size, GFP_KERNEL); - if (!txq->sw_tx_ring) { - DP_NOTICE(edev, "Tx buffers ring allocation failed\n"); - goto err; + if (txq->is_xdp) { + size = sizeof(*txq->sw_tx_ring.pages) * TX_RING_SIZE; + txq->sw_tx_ring.pages = kzalloc(size, GFP_KERNEL); + if (!txq->sw_tx_ring.pages) + goto err; + } else { + size = sizeof(*txq->sw_tx_ring.skbs) * TX_RING_SIZE; + txq->sw_tx_ring.skbs = kzalloc(size, GFP_KERNEL); + if (!txq->sw_tx_ring.skbs) + goto err; } rc = edev->ops->common->chain_alloc(edev->cdev, @@ -2966,16 +3281,13 @@ err: /* This function frees all memory of a single fp */ static void qede_free_mem_fp(struct qede_dev *edev, struct qede_fastpath *fp) { - int tc; - qede_free_mem_sb(edev, fp->sb_info); if (fp->type & QEDE_FASTPATH_RX) qede_free_mem_rxq(edev, fp->rxq); if (fp->type & QEDE_FASTPATH_TX) - for (tc = 0; tc < edev->num_tc; tc++) - qede_free_mem_txq(edev, &fp->txqs[tc]); + qede_free_mem_txq(edev, fp->txq); } /* This function allocates all memory needed for a single fp (i.e. an entity @@ -2983,28 +3295,31 @@ static void qede_free_mem_fp(struct qede_dev *edev, struct qede_fastpath *fp) */ static int qede_alloc_mem_fp(struct qede_dev *edev, struct qede_fastpath *fp) { - int rc, tc; + int rc = 0; rc = qede_alloc_mem_sb(edev, fp->sb_info, fp->id); if (rc) - goto err; + goto out; if (fp->type & QEDE_FASTPATH_RX) { rc = qede_alloc_mem_rxq(edev, fp->rxq); if (rc) - goto err; + goto out; + } + + if (fp->type & QEDE_FASTPATH_XDP) { + rc = qede_alloc_mem_txq(edev, fp->xdp_tx); + if (rc) + goto out; } if (fp->type & QEDE_FASTPATH_TX) { - for (tc = 0; tc < edev->num_tc; tc++) { - rc = qede_alloc_mem_txq(edev, &fp->txqs[tc]); - if (rc) - goto err; - } + rc = qede_alloc_mem_txq(edev, fp->txq); + if (rc) + goto out; } - return 0; -err: +out: return rc; } @@ -3043,7 +3358,7 @@ static int qede_alloc_mem_load(struct qede_dev *edev) /* This function inits fp content and resets the SB, RXQ and TXQ structures */ static void qede_init_fp(struct qede_dev *edev) { - int queue_id, rxq_index = 0, txq_index = 0, tc; + int queue_id, rxq_index = 0, txq_index = 0; struct qede_fastpath *fp; for_each_queue(queue_id) { @@ -3052,25 +3367,28 @@ static void qede_init_fp(struct qede_dev *edev) fp->edev = edev; fp->id = queue_id; - memset((void *)&fp->napi, 0, sizeof(fp->napi)); - - memset((void *)fp->sb_info, 0, sizeof(*fp->sb_info)); + if (fp->type & QEDE_FASTPATH_XDP) { + fp->xdp_tx->index = QEDE_TXQ_IDX_TO_XDP(edev, + rxq_index); + fp->xdp_tx->is_xdp = 1; + } if (fp->type & QEDE_FASTPATH_RX) { - memset((void *)fp->rxq, 0, sizeof(*fp->rxq)); fp->rxq->rxq_id = rxq_index++; + + /* Determine how to map buffers for this queue */ + if (fp->type & QEDE_FASTPATH_XDP) + fp->rxq->data_direction = DMA_BIDIRECTIONAL; + else + fp->rxq->data_direction = DMA_FROM_DEVICE; + fp->rxq->dev = &edev->pdev->dev; } if (fp->type & QEDE_FASTPATH_TX) { - memset((void *)fp->txqs, 0, - (edev->num_tc * sizeof(*fp->txqs))); - for (tc = 0; tc < edev->num_tc; tc++) { - fp->txqs[tc].index = txq_index + - tc * QEDE_TSS_COUNT(edev); - if (edev->dev_info.is_legacy) - fp->txqs[tc].is_legacy = true; - } - txq_index++; + fp->txq->index = txq_index++; + if (edev->dev_info.is_legacy) + fp->txq->is_legacy = 1; + fp->txq->dev = &edev->pdev->dev; } snprintf(fp->name, sizeof(fp->name), "%s-fp-%d", @@ -3238,11 +3556,18 @@ static int qede_drain_txq(struct qede_dev *edev, return 0; } +static int qede_stop_txq(struct qede_dev *edev, + struct qede_tx_queue *txq, int rss_id) +{ + return edev->ops->q_tx_stop(edev->cdev, rss_id, txq->handle); +} + static int qede_stop_queues(struct qede_dev *edev) { struct qed_update_vport_params vport_update_params; struct qed_dev *cdev = edev->cdev; - int rc, tc, i; + struct qede_fastpath *fp; + int rc, i; /* Disable the vport */ memset(&vport_update_params, 0, sizeof(vport_update_params)); @@ -3259,53 +3584,49 @@ static int qede_stop_queues(struct qede_dev *edev) /* Flush Tx queues. If needed, request drain from MCP */ for_each_queue(i) { - struct qede_fastpath *fp = &edev->fp_array[i]; + fp = &edev->fp_array[i]; if (fp->type & QEDE_FASTPATH_TX) { - for (tc = 0; tc < edev->num_tc; tc++) { - struct qede_tx_queue *txq = &fp->txqs[tc]; + rc = qede_drain_txq(edev, fp->txq, true); + if (rc) + return rc; + } - rc = qede_drain_txq(edev, txq, true); - if (rc) - return rc; - } + if (fp->type & QEDE_FASTPATH_XDP) { + rc = qede_drain_txq(edev, fp->xdp_tx, true); + if (rc) + return rc; } } /* Stop all Queues in reverse order */ for (i = QEDE_QUEUE_CNT(edev) - 1; i >= 0; i--) { - struct qed_stop_rxq_params rx_params; + fp = &edev->fp_array[i]; /* Stop the Tx Queue(s) */ - if (edev->fp_array[i].type & QEDE_FASTPATH_TX) { - for (tc = 0; tc < edev->num_tc; tc++) { - struct qed_stop_txq_params tx_params; - u8 val; - - tx_params.rss_id = i; - val = edev->fp_array[i].txqs[tc].index; - tx_params.tx_queue_id = val; - rc = edev->ops->q_tx_stop(cdev, &tx_params); - if (rc) { - DP_ERR(edev, "Failed to stop TXQ #%d\n", - tx_params.tx_queue_id); - return rc; - } - } + if (fp->type & QEDE_FASTPATH_TX) { + rc = qede_stop_txq(edev, fp->txq, i); + if (rc) + return rc; } /* Stop the Rx Queue */ - if (edev->fp_array[i].type & QEDE_FASTPATH_RX) { - memset(&rx_params, 0, sizeof(rx_params)); - rx_params.rss_id = i; - rx_params.rx_queue_id = edev->fp_array[i].rxq->rxq_id; - - rc = edev->ops->q_rx_stop(cdev, &rx_params); + if (fp->type & QEDE_FASTPATH_RX) { + rc = edev->ops->q_rx_stop(cdev, i, fp->rxq->handle); if (rc) { DP_ERR(edev, "Failed to stop RXQ #%d\n", i); return rc; } } + + /* Stop the XDP forwarding queue */ + if (fp->type & QEDE_FASTPATH_XDP) { + rc = qede_stop_txq(edev, fp->xdp_tx, i); + if (rc) + return rc; + + bpf_prog_put(fp->rxq->xdp_prog); + } } /* Stop the vport */ @@ -3316,9 +3637,55 @@ static int qede_stop_queues(struct qede_dev *edev) return rc; } +static int qede_start_txq(struct qede_dev *edev, + struct qede_fastpath *fp, + struct qede_tx_queue *txq, u8 rss_id, u16 sb_idx) +{ + dma_addr_t phys_table = qed_chain_get_pbl_phys(&txq->tx_pbl); + u32 page_cnt = qed_chain_get_page_cnt(&txq->tx_pbl); + struct qed_queue_start_common_params params; + struct qed_txq_start_ret_params ret_params; + int rc; + + memset(¶ms, 0, sizeof(params)); + memset(&ret_params, 0, sizeof(ret_params)); + + /* Let the XDP queue share the queue-zone with one of the regular txq. + * We don't really care about its coalescing. + */ + if (txq->is_xdp) + params.queue_id = QEDE_TXQ_XDP_TO_IDX(edev, txq); + else + params.queue_id = txq->index; + + params.sb = fp->sb_info->igu_sb_id; + params.sb_idx = sb_idx; + + rc = edev->ops->q_tx_start(edev->cdev, rss_id, ¶ms, phys_table, + page_cnt, &ret_params); + if (rc) { + DP_ERR(edev, "Start TXQ #%d failed %d\n", txq->index, rc); + return rc; + } + + txq->doorbell_addr = ret_params.p_doorbell; + txq->handle = ret_params.p_handle; + + /* Determine the FW consumer address associated */ + txq->hw_cons_ptr = &fp->sb_info->sb_virt->pi_array[sb_idx]; + + /* Prepare the doorbell parameters */ + SET_FIELD(txq->tx_db.data.params, ETH_DB_DATA_DEST, DB_DEST_XCM); + SET_FIELD(txq->tx_db.data.params, ETH_DB_DATA_AGG_CMD, DB_AGG_CMD_SET); + SET_FIELD(txq->tx_db.data.params, ETH_DB_DATA_AGG_VAL_SEL, + DQ_XCM_ETH_TX_BD_PROD_CMD); + txq->tx_db.data.agg_flags = DQ_XCM_ETH_DQ_CF_CMD; + + return rc; +} + static int qede_start_queues(struct qede_dev *edev, bool clear_stats) { - int rc, tc, i; int vlan_removal_en = 1; struct qed_dev *cdev = edev->cdev; struct qed_update_vport_params vport_update_params; @@ -3326,6 +3693,7 @@ static int qede_start_queues(struct qede_dev *edev, bool clear_stats) struct qed_dev_info *qed_info = &edev->dev_info.common; struct qed_start_vport_params start = {0}; bool reset_rss_indir = false; + int rc, i; if (!edev->num_queues) { DP_ERR(edev, @@ -3357,11 +3725,12 @@ static int qede_start_queues(struct qede_dev *edev, bool clear_stats) u32 page_cnt; if (fp->type & QEDE_FASTPATH_RX) { + struct qed_rxq_start_ret_params ret_params; struct qede_rx_queue *rxq = fp->rxq; __le16 *val; + memset(&ret_params, 0, sizeof(ret_params)); memset(&q_params, 0, sizeof(q_params)); - q_params.rss_id = i; q_params.queue_id = rxq->rxq_id; q_params.vport_id = 0; q_params.sb = fp->sb_info->igu_sb_id; @@ -3371,60 +3740,44 @@ static int qede_start_queues(struct qede_dev *edev, bool clear_stats) qed_chain_get_pbl_phys(&rxq->rx_comp_ring); page_cnt = qed_chain_get_page_cnt(&rxq->rx_comp_ring); - rc = edev->ops->q_rx_start(cdev, &q_params, + rc = edev->ops->q_rx_start(cdev, i, &q_params, rxq->rx_buf_size, rxq->rx_bd_ring.p_phys_addr, p_phys_table, - page_cnt, - &rxq->hw_rxq_prod_addr); + page_cnt, &ret_params); if (rc) { DP_ERR(edev, "Start RXQ #%d failed %d\n", i, rc); return rc; } + /* Use the return parameters */ + rxq->hw_rxq_prod_addr = ret_params.p_prod; + rxq->handle = ret_params.p_handle; + val = &fp->sb_info->sb_virt->pi_array[RX_PI]; rxq->hw_cons_ptr = val; qede_update_rx_prod(edev, rxq); } - if (!(fp->type & QEDE_FASTPATH_TX)) - continue; - - for (tc = 0; tc < edev->num_tc; tc++) { - struct qede_tx_queue *txq = &fp->txqs[tc]; - - p_phys_table = qed_chain_get_pbl_phys(&txq->tx_pbl); - page_cnt = qed_chain_get_page_cnt(&txq->tx_pbl); - - memset(&q_params, 0, sizeof(q_params)); - q_params.rss_id = i; - q_params.queue_id = txq->index; - q_params.vport_id = 0; - q_params.sb = fp->sb_info->igu_sb_id; - q_params.sb_idx = TX_PI(tc); + if (fp->type & QEDE_FASTPATH_XDP) { + rc = qede_start_txq(edev, fp, fp->xdp_tx, i, XDP_PI); + if (rc) + return rc; - rc = edev->ops->q_tx_start(cdev, &q_params, - p_phys_table, page_cnt, - &txq->doorbell_addr); - if (rc) { - DP_ERR(edev, "Start TXQ #%d failed %d\n", - txq->index, rc); + fp->rxq->xdp_prog = bpf_prog_add(edev->xdp_prog, 1); + if (IS_ERR(fp->rxq->xdp_prog)) { + rc = PTR_ERR(fp->rxq->xdp_prog); + fp->rxq->xdp_prog = NULL; return rc; } + } - txq->hw_cons_ptr = - &fp->sb_info->sb_virt->pi_array[TX_PI(tc)]; - SET_FIELD(txq->tx_db.data.params, - ETH_DB_DATA_DEST, DB_DEST_XCM); - SET_FIELD(txq->tx_db.data.params, ETH_DB_DATA_AGG_CMD, - DB_AGG_CMD_SET); - SET_FIELD(txq->tx_db.data.params, - ETH_DB_DATA_AGG_VAL_SEL, - DQ_XCM_ETH_TX_BD_PROD_CMD); - - txq->tx_db.data.agg_flags = DQ_XCM_ETH_DQ_CF_CMD; + if (fp->type & QEDE_FASTPATH_TX) { + rc = qede_start_txq(edev, fp, fp->txq, i, TX_PI(0)); + if (rc) + return rc; } } @@ -3519,15 +3872,18 @@ enum qede_unload_mode { QEDE_UNLOAD_NORMAL, }; -static void qede_unload(struct qede_dev *edev, enum qede_unload_mode mode) +static void qede_unload(struct qede_dev *edev, enum qede_unload_mode mode, + bool is_locked) { struct qed_link_params link_params; int rc; DP_INFO(edev, "Starting qede unload\n"); + if (!is_locked) + __qede_lock(edev); + qede_roce_dev_event_close(edev); - mutex_lock(&edev->qede_lock); edev->state = QEDE_STATE_CLOSED; /* Close OS Tx */ @@ -3559,7 +3915,8 @@ static void qede_unload(struct qede_dev *edev, enum qede_unload_mode mode) qede_free_fp_array(edev); out: - mutex_unlock(&edev->qede_lock); + if (!is_locked) + __qede_unlock(edev); DP_INFO(edev, "Ending qede unload\n"); } @@ -3568,7 +3925,8 @@ enum qede_load_mode { QEDE_LOAD_RELOAD, }; -static int qede_load(struct qede_dev *edev, enum qede_load_mode mode) +static int qede_load(struct qede_dev *edev, enum qede_load_mode mode, + bool is_locked) { struct qed_link_params link_params; struct qed_link_output link_output; @@ -3576,21 +3934,24 @@ static int qede_load(struct qede_dev *edev, enum qede_load_mode mode) DP_INFO(edev, "Starting qede load\n"); + if (!is_locked) + __qede_lock(edev); + rc = qede_set_num_queues(edev); if (rc) - goto err0; + goto out; rc = qede_alloc_fp_array(edev); if (rc) - goto err0; + goto out; qede_init_fp(edev); rc = qede_alloc_mem_load(edev); if (rc) goto err1; - DP_INFO(edev, "Allocated %d RSS queues on %d TC/s\n", - QEDE_QUEUE_CNT(edev), edev->num_tc); + DP_INFO(edev, "Allocated %d Rx, %d Tx queues\n", + QEDE_RSS_COUNT(edev), QEDE_TSS_COUNT(edev)); rc = qede_set_real_num_queues(edev); if (rc) @@ -3612,10 +3973,6 @@ static int qede_load(struct qede_dev *edev, enum qede_load_mode mode) /* Add primary mac and set Rx filters */ ether_addr_copy(edev->primary_mac, edev->ndev->dev_addr); - mutex_lock(&edev->qede_lock); - edev->state = QEDE_STATE_OPEN; - mutex_unlock(&edev->qede_lock); - /* Program un-configured VLANs */ qede_configure_vlan_filters(edev); @@ -3630,10 +3987,12 @@ static int qede_load(struct qede_dev *edev, enum qede_load_mode mode) qede_roce_dev_event_open(edev); qede_link_update(edev, &link_output); + edev->state = QEDE_STATE_OPEN; + DP_INFO(edev, "Ending successfully qede load\n"); - return 0; + goto out; err4: qede_sync_free_irqs(edev); memset(&edev->int_info.msix_cnt, 0, sizeof(struct qed_int_info)); @@ -3647,26 +4006,40 @@ err1: edev->num_queues = 0; edev->fp_num_tx = 0; edev->fp_num_rx = 0; -err0: +out: + if (!is_locked) + __qede_unlock(edev); + return rc; } +/* 'func' should be able to run between unload and reload assuming interface + * is actually running, or afterwards in case it's currently DOWN. + */ void qede_reload(struct qede_dev *edev, - void (*func)(struct qede_dev *, union qede_reload_args *), - union qede_reload_args *args) + struct qede_reload_args *args, bool is_locked) { - qede_unload(edev, QEDE_UNLOAD_NORMAL); - /* Call function handler to update parameters - * needed for function load. - */ - if (func) - func(edev, args); + if (!is_locked) + __qede_lock(edev); - qede_load(edev, QEDE_LOAD_RELOAD); + /* Since qede_lock is held, internal state wouldn't change even + * if netdev state would start transitioning. Check whether current + * internal configuration indicates device is up, then reload. + */ + if (edev->state == QEDE_STATE_OPEN) { + qede_unload(edev, QEDE_UNLOAD_NORMAL, true); + if (args) + args->func(edev, args); + qede_load(edev, QEDE_LOAD_RELOAD, true); + + /* Since no one is going to do it for us, re-configure */ + qede_config_rx_mode(edev->ndev); + } else if (args) { + args->func(edev, args); + } - mutex_lock(&edev->qede_lock); - qede_config_rx_mode(edev->ndev); - mutex_unlock(&edev->qede_lock); + if (!is_locked) + __qede_unlock(edev); } /* called with rtnl_lock */ @@ -3679,13 +4052,14 @@ static int qede_open(struct net_device *ndev) edev->ops->common->set_power_state(edev->cdev, PCI_D0); - rc = qede_load(edev, QEDE_LOAD_NORMAL); - + rc = qede_load(edev, QEDE_LOAD_NORMAL, false); if (rc) return rc; udp_tunnel_get_rx_info(ndev); + edev->ops->common->update_drv_state(edev->cdev, true); + return 0; } @@ -3693,7 +4067,9 @@ static int qede_close(struct net_device *ndev) { struct qede_dev *edev = netdev_priv(ndev); - qede_unload(edev, QEDE_UNLOAD_NORMAL); + qede_unload(edev, QEDE_UNLOAD_NORMAL, false); + + edev->ops->common->update_drv_state(edev->cdev, false); return 0; } @@ -3755,6 +4131,8 @@ static int qede_set_mac_addr(struct net_device *ndev, void *p) if (rc) return rc; + edev->ops->common->update_mac(edev->cdev, addr->sa_data); + /* Add MAC filter according to the new unicast HW MAC address */ ether_addr_copy(edev->primary_mac, ndev->dev_addr); return qede_set_ucast_rx_mac(edev, QED_FILTER_XCAST_TYPE_ADD, @@ -3821,15 +4199,8 @@ static void qede_set_rx_mode(struct net_device *ndev) { struct qede_dev *edev = netdev_priv(ndev); - DP_INFO(edev, "qede_set_rx_mode called\n"); - - if (edev->state != QEDE_STATE_OPEN) { - DP_INFO(edev, - "qede_set_rx_mode called while interface is down\n"); - } else { - set_bit(QEDE_SP_RX_MODE, &edev->sp_flags); - schedule_delayed_work(&edev->sp_task, 0); - } + set_bit(QEDE_SP_RX_MODE, &edev->sp_flags); + schedule_delayed_work(&edev->sp_task, 0); } /* Must be called with qede_lock held */ @@ -3877,7 +4248,7 @@ static void qede_config_rx_mode(struct net_device *ndev) /* Check for promiscuous */ if ((ndev->flags & IFF_PROMISC) || - (uc_count > 15)) { /* @@@TBD resource allocation - 1 */ + (uc_count > edev->dev_info.num_mac_filters - 1)) { accept_flags = QED_FILTER_RX_MODE_TYPE_PROMISC; } else { /* Add MAC filters according to the unicast secondary macs */ |