diff options
Diffstat (limited to 'drivers/net/ethernet/intel/i40e/i40e_txrx.c')
| -rw-r--r-- | drivers/net/ethernet/intel/i40e/i40e_txrx.c | 434 |
1 files changed, 404 insertions, 30 deletions
diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.c b/drivers/net/ethernet/intel/i40e/i40e_txrx.c index d4bb482b1a7..0f5d96ad281 100644 --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.c +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.c @@ -25,6 +25,7 @@ ******************************************************************************/ #include "i40e.h" +#include "i40e_prototype.h" static inline __le64 build_ctob(u32 td_cmd, u32 td_offset, unsigned int size, u32 td_tag) @@ -39,11 +40,12 @@ static inline __le64 build_ctob(u32 td_cmd, u32 td_offset, unsigned int size, #define I40E_TXD_CMD (I40E_TX_DESC_CMD_EOP | I40E_TX_DESC_CMD_RS) /** * i40e_program_fdir_filter - Program a Flow Director filter - * @fdir_input: Packet data that will be filter parameters + * @fdir_data: Packet data that will be filter parameters + * @raw_packet: the pre-allocated packet buffer for FDir * @pf: The pf pointer * @add: True for add/update, False for remove **/ -int i40e_program_fdir_filter(struct i40e_fdir_data *fdir_data, +int i40e_program_fdir_filter(struct i40e_fdir_filter *fdir_data, u8 *raw_packet, struct i40e_pf *pf, bool add) { struct i40e_filter_program_desc *fdir_desc; @@ -68,8 +70,8 @@ int i40e_program_fdir_filter(struct i40e_fdir_data *fdir_data, tx_ring = vsi->tx_rings[0]; dev = tx_ring->dev; - dma = dma_map_single(dev, fdir_data->raw_packet, - I40E_FDIR_MAX_RAW_PACKET_LOOKUP, DMA_TO_DEVICE); + dma = dma_map_single(dev, raw_packet, + I40E_FDIR_MAX_RAW_PACKET_SIZE, DMA_TO_DEVICE); if (dma_mapping_error(dev, dma)) goto dma_fail; @@ -132,14 +134,14 @@ int i40e_program_fdir_filter(struct i40e_fdir_data *fdir_data, tx_ring->next_to_use = (i + 1 < tx_ring->count) ? i + 1 : 0; /* record length, and DMA address */ - dma_unmap_len_set(tx_buf, len, I40E_FDIR_MAX_RAW_PACKET_LOOKUP); + dma_unmap_len_set(tx_buf, len, I40E_FDIR_MAX_RAW_PACKET_SIZE); dma_unmap_addr_set(tx_buf, dma, dma); tx_desc->buffer_addr = cpu_to_le64(dma); td_cmd = I40E_TXD_CMD | I40E_TX_DESC_CMD_DUMMY; tx_desc->cmd_type_offset_bsz = - build_ctob(td_cmd, 0, I40E_FDIR_MAX_RAW_PACKET_LOOKUP, 0); + build_ctob(td_cmd, 0, I40E_FDIR_MAX_RAW_PACKET_SIZE, 0); /* set the timestamp */ tx_buf->time_stamp = jiffies; @@ -161,26 +163,329 @@ dma_fail: return -1; } +#define IP_HEADER_OFFSET 14 +#define I40E_UDPIP_DUMMY_PACKET_LEN 42 +/** + * i40e_add_del_fdir_udpv4 - Add/Remove UDPv4 filters + * @vsi: pointer to the targeted VSI + * @fd_data: the flow director data required for the FDir descriptor + * @raw_packet: the pre-allocated packet buffer for FDir + * @add: true adds a filter, false removes it + * + * Returns 0 if the filters were successfully added or removed + **/ +static int i40e_add_del_fdir_udpv4(struct i40e_vsi *vsi, + struct i40e_fdir_filter *fd_data, + u8 *raw_packet, bool add) +{ + struct i40e_pf *pf = vsi->back; + struct udphdr *udp; + struct iphdr *ip; + bool err = false; + int ret; + int i; + static char packet[] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x08, 0, + 0x45, 0, 0, 0x1c, 0, 0, 0x40, 0, 0x40, 0x11, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}; + + memcpy(raw_packet, packet, I40E_UDPIP_DUMMY_PACKET_LEN); + + ip = (struct iphdr *)(raw_packet + IP_HEADER_OFFSET); + udp = (struct udphdr *)(raw_packet + IP_HEADER_OFFSET + + sizeof(struct iphdr)); + + ip->daddr = fd_data->dst_ip[0]; + udp->dest = fd_data->dst_port; + ip->saddr = fd_data->src_ip[0]; + udp->source = fd_data->src_port; + + for (i = I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP; + i <= I40E_FILTER_PCTYPE_NONF_IPV4_UDP; i++) { + fd_data->pctype = i; + ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add); + + if (ret) { + dev_info(&pf->pdev->dev, + "Filter command send failed for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + err = true; + } else { + dev_info(&pf->pdev->dev, + "Filter OK for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + } + } + + return err ? -EOPNOTSUPP : 0; +} + +#define I40E_TCPIP_DUMMY_PACKET_LEN 54 +/** + * i40e_add_del_fdir_tcpv4 - Add/Remove TCPv4 filters + * @vsi: pointer to the targeted VSI + * @fd_data: the flow director data required for the FDir descriptor + * @raw_packet: the pre-allocated packet buffer for FDir + * @add: true adds a filter, false removes it + * + * Returns 0 if the filters were successfully added or removed + **/ +static int i40e_add_del_fdir_tcpv4(struct i40e_vsi *vsi, + struct i40e_fdir_filter *fd_data, + u8 *raw_packet, bool add) +{ + struct i40e_pf *pf = vsi->back; + struct tcphdr *tcp; + struct iphdr *ip; + bool err = false; + int ret; + /* Dummy packet */ + static char packet[] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x08, 0, + 0x45, 0, 0, 0x28, 0, 0, 0x40, 0, 0x40, 0x6, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x80, 0x11, + 0x0, 0x72, 0, 0, 0, 0}; + + memcpy(raw_packet, packet, I40E_TCPIP_DUMMY_PACKET_LEN); + + ip = (struct iphdr *)(raw_packet + IP_HEADER_OFFSET); + tcp = (struct tcphdr *)(raw_packet + IP_HEADER_OFFSET + + sizeof(struct iphdr)); + + ip->daddr = fd_data->dst_ip[0]; + tcp->dest = fd_data->dst_port; + ip->saddr = fd_data->src_ip[0]; + tcp->source = fd_data->src_port; + + if (add) { + if (pf->flags & I40E_FLAG_FD_ATR_ENABLED) { + dev_info(&pf->pdev->dev, "Forcing ATR off, sideband rules for TCP/IPv4 flow being applied\n"); + pf->flags &= ~I40E_FLAG_FD_ATR_ENABLED; + } + } + + fd_data->pctype = I40E_FILTER_PCTYPE_NONF_IPV4_TCP_SYN; + ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add); + + if (ret) { + dev_info(&pf->pdev->dev, + "Filter command send failed for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + err = true; + } else { + dev_info(&pf->pdev->dev, "Filter OK for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + } + + fd_data->pctype = I40E_FILTER_PCTYPE_NONF_IPV4_TCP; + + ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add); + if (ret) { + dev_info(&pf->pdev->dev, + "Filter command send failed for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + err = true; + } else { + dev_info(&pf->pdev->dev, "Filter OK for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + } + + return err ? -EOPNOTSUPP : 0; +} + +/** + * i40e_add_del_fdir_sctpv4 - Add/Remove SCTPv4 Flow Director filters for + * a specific flow spec + * @vsi: pointer to the targeted VSI + * @fd_data: the flow director data required for the FDir descriptor + * @raw_packet: the pre-allocated packet buffer for FDir + * @add: true adds a filter, false removes it + * + * Always returns -EOPNOTSUPP + **/ +static int i40e_add_del_fdir_sctpv4(struct i40e_vsi *vsi, + struct i40e_fdir_filter *fd_data, + u8 *raw_packet, bool add) +{ + return -EOPNOTSUPP; +} + +#define I40E_IP_DUMMY_PACKET_LEN 34 +/** + * i40e_add_del_fdir_ipv4 - Add/Remove IPv4 Flow Director filters for + * a specific flow spec + * @vsi: pointer to the targeted VSI + * @fd_data: the flow director data required for the FDir descriptor + * @raw_packet: the pre-allocated packet buffer for FDir + * @add: true adds a filter, false removes it + * + * Returns 0 if the filters were successfully added or removed + **/ +static int i40e_add_del_fdir_ipv4(struct i40e_vsi *vsi, + struct i40e_fdir_filter *fd_data, + u8 *raw_packet, bool add) +{ + struct i40e_pf *pf = vsi->back; + struct iphdr *ip; + bool err = false; + int ret; + int i; + static char packet[] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x08, 0, + 0x45, 0, 0, 0x14, 0, 0, 0x40, 0, 0x40, 0x10, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0}; + + memcpy(raw_packet, packet, I40E_IP_DUMMY_PACKET_LEN); + ip = (struct iphdr *)(raw_packet + IP_HEADER_OFFSET); + + ip->saddr = fd_data->src_ip[0]; + ip->daddr = fd_data->dst_ip[0]; + ip->protocol = 0; + + for (i = I40E_FILTER_PCTYPE_NONF_IPV4_OTHER; + i <= I40E_FILTER_PCTYPE_FRAG_IPV4; i++) { + fd_data->pctype = i; + ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add); + + if (ret) { + dev_info(&pf->pdev->dev, + "Filter command send failed for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + err = true; + } else { + dev_info(&pf->pdev->dev, + "Filter OK for PCTYPE %d (ret = %d)\n", + fd_data->pctype, ret); + } + } + + return err ? -EOPNOTSUPP : 0; +} + +/** + * i40e_add_del_fdir - Build raw packets to add/del fdir filter + * @vsi: pointer to the targeted VSI + * @cmd: command to get or set RX flow classification rules + * @add: true adds a filter, false removes it + * + **/ +int i40e_add_del_fdir(struct i40e_vsi *vsi, + struct i40e_fdir_filter *input, bool add) +{ + struct i40e_pf *pf = vsi->back; + u8 *raw_packet; + int ret; + + /* Populate the Flow Director that we have at the moment + * and allocate the raw packet buffer for the calling functions + */ + raw_packet = kzalloc(I40E_FDIR_MAX_RAW_PACKET_SIZE, GFP_KERNEL); + if (!raw_packet) + return -ENOMEM; + + switch (input->flow_type & ~FLOW_EXT) { + case TCP_V4_FLOW: + ret = i40e_add_del_fdir_tcpv4(vsi, input, raw_packet, + add); + break; + case UDP_V4_FLOW: + ret = i40e_add_del_fdir_udpv4(vsi, input, raw_packet, + add); + break; + case SCTP_V4_FLOW: + ret = i40e_add_del_fdir_sctpv4(vsi, input, raw_packet, + add); + break; + case IPV4_FLOW: + ret = i40e_add_del_fdir_ipv4(vsi, input, raw_packet, + add); + break; + case IP_USER_FLOW: + switch (input->ip4_proto) { + case IPPROTO_TCP: + ret = i40e_add_del_fdir_tcpv4(vsi, input, + raw_packet, add); + break; + case IPPROTO_UDP: + ret = i40e_add_del_fdir_udpv4(vsi, input, + raw_packet, add); + break; + case IPPROTO_SCTP: + ret = i40e_add_del_fdir_sctpv4(vsi, input, + raw_packet, add); + break; + default: + ret = i40e_add_del_fdir_ipv4(vsi, input, + raw_packet, add); + break; + } + break; + default: + dev_info(&pf->pdev->dev, "Could not specify spec type %d", + input->flow_type); + ret = -EINVAL; + } + + kfree(raw_packet); + return ret; +} + /** * i40e_fd_handle_status - check the Programming Status for FD * @rx_ring: the Rx ring for this descriptor - * @qw: the descriptor data + * @rx_desc: the Rx descriptor for programming Status, not a packet descriptor. * @prog_id: the id originally used for programming * * This is used to verify if the FD programming or invalidation * requested by SW to the HW is successful or not and take actions accordingly. **/ -static void i40e_fd_handle_status(struct i40e_ring *rx_ring, u32 qw, u8 prog_id) +static void i40e_fd_handle_status(struct i40e_ring *rx_ring, + union i40e_rx_desc *rx_desc, u8 prog_id) { - struct pci_dev *pdev = rx_ring->vsi->back->pdev; + struct i40e_pf *pf = rx_ring->vsi->back; + struct pci_dev *pdev = pf->pdev; + u32 fcnt_prog, fcnt_avail; u32 error; + u64 qw; + qw = le64_to_cpu(rx_desc->wb.qword1.status_error_len); error = (qw & I40E_RX_PROG_STATUS_DESC_QW1_ERROR_MASK) >> I40E_RX_PROG_STATUS_DESC_QW1_ERROR_SHIFT; - /* for now just print the Status */ - dev_info(&pdev->dev, "FD programming id %02x, Status %08x\n", - prog_id, error); + if (error == (0x1 << I40E_RX_PROG_STATUS_DESC_FD_TBL_FULL_SHIFT)) { + dev_warn(&pdev->dev, "ntuple filter loc = %d, could not be added\n", + rx_desc->wb.qword0.hi_dword.fd_id); + + /* filter programming failed most likely due to table full */ + fcnt_prog = i40e_get_current_fd_count(pf); + fcnt_avail = pf->hw.fdir_shared_filter_count + + pf->fdir_pf_filter_count; + + /* If ATR is running fcnt_prog can quickly change, + * if we are very close to full, it makes sense to disable + * FD ATR/SB and then re-enable it when there is room. + */ + if (fcnt_prog >= (fcnt_avail - I40E_FDIR_BUFFER_FULL_MARGIN)) { + /* Turn off ATR first */ + if (pf->flags | I40E_FLAG_FD_ATR_ENABLED) { + pf->flags &= ~I40E_FLAG_FD_ATR_ENABLED; + dev_warn(&pdev->dev, "FD filter space full, ATR for further flows will be turned off\n"); + pf->auto_disable_flags |= + I40E_FLAG_FD_ATR_ENABLED; + pf->flags |= I40E_FLAG_FDIR_REQUIRES_REINIT; + } else if (pf->flags | I40E_FLAG_FD_SB_ENABLED) { + pf->flags &= ~I40E_FLAG_FD_SB_ENABLED; + dev_warn(&pdev->dev, "FD filter space full, new ntuple rules will not be added\n"); + pf->auto_disable_flags |= + I40E_FLAG_FD_SB_ENABLED; + pf->flags |= I40E_FLAG_FDIR_REQUIRES_REINIT; + } + } else { + dev_info(&pdev->dev, "FD filter programming error"); + } + } else if (error == + (0x1 << I40E_RX_PROG_STATUS_DESC_NO_FD_ENTRY_SHIFT)) { + if (I40E_DEBUG_FD & pf->hw.debug_mask) + dev_info(&pdev->dev, "ntuple filter loc = %d, could not be removed\n", + rx_desc->wb.qword0.hi_dword.fd_id); + } } /** @@ -315,6 +620,20 @@ static bool i40e_check_tx_hang(struct i40e_ring *tx_ring) } /** + * i40e_get_head - Retrieve head from head writeback + * @tx_ring: tx ring to fetch head of + * + * Returns value of Tx ring head based on value stored + * in head write-back location + **/ +static inline u32 i40e_get_head(struct i40e_ring *tx_ring) +{ + void *head = (struct i40e_tx_desc *)tx_ring->desc + tx_ring->count; + + return le32_to_cpu(*(volatile __le32 *)head); +} + +/** * i40e_clean_tx_irq - Reclaim resources after transmit completes * @tx_ring: tx ring to clean * @budget: how many cleans we're allowed @@ -325,6 +644,7 @@ static bool i40e_clean_tx_irq(struct i40e_ring *tx_ring, int budget) { u16 i = tx_ring->next_to_clean; struct i40e_tx_buffer *tx_buf; + struct i40e_tx_desc *tx_head; struct i40e_tx_desc *tx_desc; unsigned int total_packets = 0; unsigned int total_bytes = 0; @@ -333,6 +653,8 @@ static bool i40e_clean_tx_irq(struct i40e_ring *tx_ring, int budget) tx_desc = I40E_TX_DESC(tx_ring, i); i -= tx_ring->count; + tx_head = I40E_TX_DESC(tx_ring, i40e_get_head(tx_ring)); + do { struct i40e_tx_desc *eop_desc = tx_buf->next_to_watch; @@ -343,9 +665,8 @@ static bool i40e_clean_tx_irq(struct i40e_ring *tx_ring, int budget) /* prevent any other reads prior to eop_desc */ read_barrier_depends(); - /* if the descriptor isn't done, no work yet to do */ - if (!(eop_desc->cmd_type_offset_bsz & - cpu_to_le64(I40E_TX_DESC_DTYPE_DESC_DONE))) + /* we have caught up to head, no work left to do */ + if (tx_head == tx_desc) break; /* clear next_to_watch to prevent false hangs */ @@ -577,7 +898,7 @@ static void i40e_clean_programming_status(struct i40e_ring *rx_ring, I40E_RX_PROG_STATUS_DESC_QW1_PROGID_SHIFT; if (id == I40E_RX_PROG_STATUS_DESC_FD_FILTER_STATUS) - i40e_fd_handle_status(rx_ring, qw, id); + i40e_fd_handle_status(rx_ring, rx_desc, id); } /** @@ -601,6 +922,10 @@ int i40e_setup_tx_descriptors(struct i40e_ring *tx_ring) /* round up to nearest 4K */ tx_ring->size = tx_ring->count * sizeof(struct i40e_tx_desc); + /* add u32 for head writeback, align after this takes care of + * guaranteeing this is at least one cache line in size + */ + tx_ring->size += sizeof(u32); tx_ring->size = ALIGN(tx_ring->size, 4096); tx_ring->desc = dma_alloc_coherent(dev, tx_ring->size, &tx_ring->dma, GFP_KERNEL); @@ -892,7 +1217,7 @@ static inline void i40e_rx_checksum(struct i40e_vsi *vsi, rx_status & (1 << I40E_RX_DESC_STATUS_L3L4P_SHIFT))) return; - /* likely incorrect csum if alternate IP extention headers found */ + /* likely incorrect csum if alternate IP extension headers found */ if (rx_status & (1 << I40E_RX_DESC_STATUS_IPV6EXADD_SHIFT)) return; @@ -956,6 +1281,29 @@ static inline u32 i40e_rx_hash(struct i40e_ring *ring, } /** + * i40e_ptype_to_hash - get a hash type + * @ptype: the ptype value from the descriptor + * + * Returns a hash type to be used by skb_set_hash + **/ +static inline enum pkt_hash_types i40e_ptype_to_hash(u8 ptype) +{ + struct i40e_rx_ptype_decoded decoded = decode_rx_desc_ptype(ptype); + + if (!decoded.known) + return PKT_HASH_TYPE_NONE; + + if (decoded.outer_ip == I40E_RX_PTYPE_OUTER_IP && + decoded.payload_layer == I40E_RX_PTYPE_PAYLOAD_LAYER_PAY4) + return PKT_HASH_TYPE_L4; + else if (decoded.outer_ip == I40E_RX_PTYPE_OUTER_IP && + decoded.payload_layer == I40E_RX_PTYPE_PAYLOAD_LAYER_PAY3) + return PKT_HASH_TYPE_L3; + else + return PKT_HASH_TYPE_L2; +} + +/** * i40e_clean_rx_irq - Reclaim resources after receive completes * @rx_ring: rx ring to clean * @budget: how many cleans we're allowed @@ -972,8 +1320,11 @@ static int i40e_clean_rx_irq(struct i40e_ring *rx_ring, int budget) u16 i = rx_ring->next_to_clean; union i40e_rx_desc *rx_desc; u32 rx_error, rx_status; + u8 rx_ptype; u64 qword; - u16 rx_ptype; + + if (budget <= 0) + return 0; rx_desc = I40E_RX_DESC(rx_ring, i); qword = le64_to_cpu(rx_desc->wb.qword1.status_error_len); @@ -1087,7 +1438,8 @@ static int i40e_clean_rx_irq(struct i40e_ring *rx_ring, int budget) goto next_desc; } - skb->rxhash = i40e_rx_hash(rx_ring, rx_desc); + skb_set_hash(skb, i40e_rx_hash(rx_ring, rx_desc), + i40e_ptype_to_hash(rx_ptype)); if (unlikely(rx_status & I40E_RXD_QW1_STATUS_TSYNVALID_MASK)) { i40e_ptp_rx_hwtstamp(vsi->back, skb, (rx_status & I40E_RXD_QW1_STATUS_TSYNINDX_MASK) >> @@ -1246,8 +1598,6 @@ static void i40e_atr(struct i40e_ring *tx_ring, struct sk_buff *skb, if (!tx_ring->atr_sample_rate) return; - tx_ring->atr_count++; - /* snag network header to get L4 type and address */ hdr.network = skb_network_header(skb); @@ -1269,8 +1619,17 @@ static void i40e_atr(struct i40e_ring *tx_ring, struct sk_buff *skb, th = (struct tcphdr *)(hdr.network + hlen); - /* sample on all syn/fin packets or once every atr sample rate */ - if (!th->fin && !th->syn && (tx_ring->atr_count < tx_ring->atr_sample_rate)) + /* Due to lack of space, no more new filters can be programmed */ + if (th->syn && (pf->auto_disable_flags & I40E_FLAG_FD_ATR_ENABLED)) + return; + + tx_ring->atr_count++; + + /* sample on all syn/fin/rst packets or once every atr sample rate */ + if (!th->fin && + !th->syn && + !th->rst && + (tx_ring->atr_count < tx_ring->atr_sample_rate)) return; tx_ring->atr_count = 0; @@ -1294,7 +1653,7 @@ static void i40e_atr(struct i40e_ring *tx_ring, struct sk_buff *skb, dtype_cmd = I40E_TX_DESC_DTYPE_FILTER_PROG; - dtype_cmd |= th->fin ? + dtype_cmd |= (th->fin || th->rst) ? (I40E_FILTER_PROGRAM_DESC_PCMD_REMOVE << I40E_TXD_FLTR_QW1_PCMD_SHIFT) : (I40E_FILTER_PROGRAM_DESC_PCMD_ADD_UPDATE << @@ -1596,7 +1955,8 @@ static void i40e_create_tx_ctx(struct i40e_ring *tx_ring, struct i40e_tx_context_desc *context_desc; int i = tx_ring->next_to_use; - if (!cd_type_cmd_tso_mss && !cd_tunneling && !cd_l2tag2) + if ((cd_type_cmd_tso_mss == I40E_TX_DESC_DTYPE_CONTEXT) && + !cd_tunneling && !cd_l2tag2) return; /* grab the next descriptor */ @@ -1707,9 +2067,23 @@ static void i40e_tx_map(struct i40e_ring *tx_ring, struct sk_buff *skb, tx_bi = &tx_ring->tx_bi[i]; } - tx_desc->cmd_type_offset_bsz = - build_ctob(td_cmd, td_offset, size, td_tag) | - cpu_to_le64((u64)I40E_TXD_CMD << I40E_TXD_QW1_CMD_SHIFT); + /* Place RS bit on last descriptor of any packet that spans across the + * 4th descriptor (WB_STRIDE aka 0x3) in a 64B cacheline. + */ +#define WB_STRIDE 0x3 + if (((i & WB_STRIDE) != WB_STRIDE) && + (first <= &tx_ring->tx_bi[i]) && + (first >= &tx_ring->tx_bi[i & ~WB_STRIDE])) { + tx_desc->cmd_type_offset_bsz = + build_ctob(td_cmd, td_offset, size, td_tag) | + cpu_to_le64((u64)I40E_TX_DESC_CMD_EOP << + I40E_TXD_QW1_CMD_SHIFT); + } else { + tx_desc->cmd_type_offset_bsz = + build_ctob(td_cmd, td_offset, size, td_tag) | + cpu_to_le64((u64)I40E_TXD_CMD << + I40E_TXD_QW1_CMD_SHIFT); + } netdev_tx_sent_queue(netdev_get_tx_queue(tx_ring->netdev, tx_ring->queue_index), @@ -1812,7 +2186,7 @@ static int i40e_xmit_descriptor_count(struct sk_buff *skb, /* need: 1 descriptor per page * PAGE_SIZE/I40E_MAX_DATA_PER_TXD, * + 1 desc for skb_head_len/I40E_MAX_DATA_PER_TXD, - * + 2 desc gap to keep tail from touching head, + * + 4 desc gap to avoid the cache line where head is, * + 1 desc for context descriptor, * otherwise try next time */ @@ -1823,7 +2197,7 @@ static int i40e_xmit_descriptor_count(struct sk_buff *skb, count += skb_shinfo(skb)->nr_frags; #endif count += TXD_USE_COUNT(skb_headlen(skb)); - if (i40e_maybe_stop_tx(tx_ring, count + 3)) { + if (i40e_maybe_stop_tx(tx_ring, count + 4 + 1)) { tx_ring->tx_stats.tx_busy++; return 0; } |