diff options
Diffstat (limited to 'drivers/net/ethernet/intel/i40evf/i40e_txrx.c')
| -rw-r--r-- | drivers/net/ethernet/intel/i40evf/i40e_txrx.c | 90 |
1 files changed, 75 insertions, 15 deletions
diff --git a/drivers/net/ethernet/intel/i40evf/i40e_txrx.c b/drivers/net/ethernet/intel/i40evf/i40e_txrx.c index ffdb01d853db..53be5f44d015 100644 --- a/drivers/net/ethernet/intel/i40evf/i40e_txrx.c +++ b/drivers/net/ethernet/intel/i40evf/i40e_txrx.c @@ -1,7 +1,7 @@ /******************************************************************************* * * Intel Ethernet Controller XL710 Family Linux Virtual Function Driver - * Copyright(c) 2013 Intel Corporation. + * Copyright(c) 2013 - 2014 Intel Corporation. * * This program is free software; you can redistribute it and/or modify it * under the terms and conditions of the GNU General Public License, @@ -24,6 +24,7 @@ #include <linux/prefetch.h> #include "i40evf.h" +#include "i40e_prototype.h" static inline __le64 build_ctob(u32 td_cmd, u32 td_offset, unsigned int size, u32 td_tag) @@ -169,6 +170,20 @@ 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 @@ -179,6 +194,7 @@ 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; @@ -187,6 +203,8 @@ 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; @@ -197,9 +215,8 @@ 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 */ @@ -431,6 +448,10 @@ int i40evf_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); @@ -722,7 +743,7 @@ 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; @@ -786,6 +807,29 @@ 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 @@ -802,13 +846,13 @@ 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; rx_desc = I40E_RX_DESC(rx_ring, i); qword = le64_to_cpu(rx_desc->wb.qword1.status_error_len); - rx_status = (qword & I40E_RXD_QW1_STATUS_MASK) - >> I40E_RXD_QW1_STATUS_SHIFT; + rx_status = (qword & I40E_RXD_QW1_STATUS_MASK) >> + I40E_RXD_QW1_STATUS_SHIFT; while (rx_status & (1 << I40E_RX_DESC_STATUS_DD_SHIFT)) { union i40e_rx_desc *next_rxd; @@ -912,7 +956,8 @@ 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)); /* probably a little skewed due to removing CRC */ total_rx_bytes += skb->len; total_rx_packets++; @@ -1241,7 +1286,8 @@ 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 */ @@ -1352,9 +1398,23 @@ 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), @@ -1457,7 +1517,7 @@ 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 */ @@ -1468,7 +1528,7 @@ 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; } |