提交 a5e9c572 编写于 作者: A Alexander Duyck 提交者: Jeff Kirsher

i40e: clean up Tx fast path

Sync the fast path for i40e_tx_map and i40e_clean_tx_irq so that they
are similar to igb and ixgbe.

- Only update the Tx descriptor ring in tx_map
- Make skb mapping always on the first buffer in the chain
- Drop the use of MAPPED_AS_PAGE Tx flag
- Only store flags on the first buffer_info structure
Signed-off-by: NAlexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NKavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
上级 fc4ac67b
......@@ -189,27 +189,30 @@ static void i40e_fd_handle_status(struct i40e_ring *rx_ring, u32 qw, u8 prog_id)
}
/**
* i40e_unmap_tx_resource - Release a Tx buffer
* i40e_unmap_and_free_tx_resource - Release a Tx buffer
* @ring: the ring that owns the buffer
* @tx_buffer: the buffer to free
**/
static inline void i40e_unmap_tx_resource(struct i40e_ring *ring,
struct i40e_tx_buffer *tx_buffer)
static void i40e_unmap_and_free_tx_resource(struct i40e_ring *ring,
struct i40e_tx_buffer *tx_buffer)
{
if (dma_unmap_len(tx_buffer, len)) {
if (tx_buffer->tx_flags & I40E_TX_FLAGS_MAPPED_AS_PAGE)
dma_unmap_page(ring->dev,
dma_unmap_addr(tx_buffer, dma),
dma_unmap_len(tx_buffer, len),
DMA_TO_DEVICE);
else
if (tx_buffer->skb) {
dev_kfree_skb_any(tx_buffer->skb);
if (dma_unmap_len(tx_buffer, len))
dma_unmap_single(ring->dev,
dma_unmap_addr(tx_buffer, dma),
dma_unmap_len(tx_buffer, len),
DMA_TO_DEVICE);
} else if (dma_unmap_len(tx_buffer, len)) {
dma_unmap_page(ring->dev,
dma_unmap_addr(tx_buffer, dma),
dma_unmap_len(tx_buffer, len),
DMA_TO_DEVICE);
}
tx_buffer->time_stamp = 0;
tx_buffer->next_to_watch = NULL;
tx_buffer->skb = NULL;
dma_unmap_len_set(tx_buffer, len, 0);
/* tx_buffer must be completely set up in the transmit path */
}
/**
......@@ -218,7 +221,6 @@ static inline void i40e_unmap_tx_resource(struct i40e_ring *ring,
**/
void i40e_clean_tx_ring(struct i40e_ring *tx_ring)
{
struct i40e_tx_buffer *tx_buffer;
unsigned long bi_size;
u16 i;
......@@ -227,13 +229,8 @@ void i40e_clean_tx_ring(struct i40e_ring *tx_ring)
return;
/* Free all the Tx ring sk_buffs */
for (i = 0; i < tx_ring->count; i++) {
tx_buffer = &tx_ring->tx_bi[i];
i40e_unmap_tx_resource(tx_ring, tx_buffer);
if (tx_buffer->skb)
dev_kfree_skb_any(tx_buffer->skb);
tx_buffer->skb = NULL;
}
for (i = 0; i < tx_ring->count; i++)
i40e_unmap_and_free_tx_resource(tx_ring, &tx_ring->tx_bi[i]);
bi_size = sizeof(struct i40e_tx_buffer) * tx_ring->count;
memset(tx_ring->tx_bi, 0, bi_size);
......@@ -332,16 +329,18 @@ static bool i40e_clean_tx_irq(struct i40e_ring *tx_ring, int budget)
tx_buf = &tx_ring->tx_bi[i];
tx_desc = I40E_TX_DESC(tx_ring, i);
i -= tx_ring->count;
for (; budget; budget--) {
struct i40e_tx_desc *eop_desc;
eop_desc = tx_buf->next_to_watch;
do {
struct i40e_tx_desc *eop_desc = tx_buf->next_to_watch;
/* if next_to_watch is not set then there is no work pending */
if (!eop_desc)
break;
/* 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)))
......@@ -349,44 +348,67 @@ static bool i40e_clean_tx_irq(struct i40e_ring *tx_ring, int budget)
/* clear next_to_watch to prevent false hangs */
tx_buf->next_to_watch = NULL;
tx_buf->time_stamp = 0;
/* set memory barrier before eop_desc is verified */
rmb();
/* update the statistics for this packet */
total_bytes += tx_buf->bytecount;
total_packets += tx_buf->gso_segs;
do {
i40e_unmap_tx_resource(tx_ring, tx_buf);
/* free the skb */
dev_kfree_skb_any(tx_buf->skb);
/* clear dtype status */
tx_desc->cmd_type_offset_bsz &=
~cpu_to_le64(I40E_TXD_QW1_DTYPE_MASK);
/* unmap skb header data */
dma_unmap_single(tx_ring->dev,
dma_unmap_addr(tx_buf, dma),
dma_unmap_len(tx_buf, len),
DMA_TO_DEVICE);
if (likely(tx_desc == eop_desc)) {
eop_desc = NULL;
/* clear tx_buffer data */
tx_buf->skb = NULL;
dma_unmap_len_set(tx_buf, len, 0);
dev_kfree_skb_any(tx_buf->skb);
tx_buf->skb = NULL;
total_bytes += tx_buf->bytecount;
total_packets += tx_buf->gso_segs;
}
/* unmap remaining buffers */
while (tx_desc != eop_desc) {
tx_buf++;
tx_desc++;
i++;
if (unlikely(i == tx_ring->count)) {
i = 0;
if (unlikely(!i)) {
i -= tx_ring->count;
tx_buf = tx_ring->tx_bi;
tx_desc = I40E_TX_DESC(tx_ring, 0);
}
} while (eop_desc);
}
/* unmap any remaining paged data */
if (dma_unmap_len(tx_buf, len)) {
dma_unmap_page(tx_ring->dev,
dma_unmap_addr(tx_buf, dma),
dma_unmap_len(tx_buf, len),
DMA_TO_DEVICE);
dma_unmap_len_set(tx_buf, len, 0);
}
}
/* move us one more past the eop_desc for start of next pkt */
tx_buf++;
tx_desc++;
i++;
if (unlikely(!i)) {
i -= tx_ring->count;
tx_buf = tx_ring->tx_bi;
tx_desc = I40E_TX_DESC(tx_ring, 0);
}
/* update budget accounting */
budget--;
} while (likely(budget));
i += tx_ring->count;
tx_ring->next_to_clean = i;
tx_ring->tx_stats.bytes += total_bytes;
tx_ring->tx_stats.packets += total_packets;
tx_ring->q_vector->tx.total_bytes += total_bytes;
tx_ring->q_vector->tx.total_packets += total_packets;
if (check_for_tx_hang(tx_ring) && i40e_check_tx_hang(tx_ring)) {
/* schedule immediate reset if we believe we hung */
dev_info(tx_ring->dev, "Detected Tx Unit Hang\n"
......@@ -1515,68 +1537,71 @@ static void i40e_tx_map(struct i40e_ring *tx_ring, struct sk_buff *skb,
struct i40e_tx_buffer *first, u32 tx_flags,
const u8 hdr_len, u32 td_cmd, u32 td_offset)
{
struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[0];
unsigned int data_len = skb->data_len;
unsigned int size = skb_headlen(skb);
struct device *dev = tx_ring->dev;
u32 paylen = skb->len - hdr_len;
u16 i = tx_ring->next_to_use;
struct skb_frag_struct *frag;
struct i40e_tx_buffer *tx_bi;
struct i40e_tx_desc *tx_desc;
u32 buf_offset = 0;
u16 i = tx_ring->next_to_use;
u32 td_tag = 0;
dma_addr_t dma;
u16 gso_segs;
dma = dma_map_single(dev, skb->data, size, DMA_TO_DEVICE);
if (dma_mapping_error(dev, dma))
goto dma_error;
if (tx_flags & I40E_TX_FLAGS_HW_VLAN) {
td_cmd |= I40E_TX_DESC_CMD_IL2TAG1;
td_tag = (tx_flags & I40E_TX_FLAGS_VLAN_MASK) >>
I40E_TX_FLAGS_VLAN_SHIFT;
}
if (tx_flags & (I40E_TX_FLAGS_TSO | I40E_TX_FLAGS_FSO))
gso_segs = skb_shinfo(skb)->gso_segs;
else
gso_segs = 1;
/* multiply data chunks by size of headers */
first->bytecount = skb->len - hdr_len + (gso_segs * hdr_len);
first->gso_segs = gso_segs;
first->skb = skb;
first->tx_flags = tx_flags;
dma = dma_map_single(tx_ring->dev, skb->data, size, DMA_TO_DEVICE);
tx_desc = I40E_TX_DESC(tx_ring, i);
for (;;) {
while (size > I40E_MAX_DATA_PER_TXD) {
tx_desc->buffer_addr = cpu_to_le64(dma + buf_offset);
tx_bi = first;
for (frag = &skb_shinfo(skb)->frags[0];; frag++) {
if (dma_mapping_error(tx_ring->dev, dma))
goto dma_error;
/* record length, and DMA address */
dma_unmap_len_set(tx_bi, len, size);
dma_unmap_addr_set(tx_bi, dma, dma);
tx_desc->buffer_addr = cpu_to_le64(dma);
while (unlikely(size > I40E_MAX_DATA_PER_TXD)) {
tx_desc->cmd_type_offset_bsz =
build_ctob(td_cmd, td_offset,
I40E_MAX_DATA_PER_TXD, td_tag);
buf_offset += I40E_MAX_DATA_PER_TXD;
size -= I40E_MAX_DATA_PER_TXD;
tx_desc++;
i++;
if (i == tx_ring->count) {
tx_desc = I40E_TX_DESC(tx_ring, 0);
i = 0;
}
}
tx_bi = &tx_ring->tx_bi[i];
dma_unmap_len_set(tx_bi, len, buf_offset + size);
dma_unmap_addr_set(tx_bi, dma, dma);
tx_bi->tx_flags = tx_flags;
dma += I40E_MAX_DATA_PER_TXD;
size -= I40E_MAX_DATA_PER_TXD;
tx_desc->buffer_addr = cpu_to_le64(dma + buf_offset);
tx_desc->cmd_type_offset_bsz = build_ctob(td_cmd, td_offset,
size, td_tag);
tx_desc->buffer_addr = cpu_to_le64(dma);
}
if (likely(!data_len))
break;
size = skb_frag_size(frag);
data_len -= size;
buf_offset = 0;
tx_flags |= I40E_TX_FLAGS_MAPPED_AS_PAGE;
dma = skb_frag_dma_map(dev, frag, 0, size, DMA_TO_DEVICE);
if (dma_mapping_error(dev, dma))
goto dma_error;
tx_desc->cmd_type_offset_bsz = build_ctob(td_cmd, td_offset,
size, td_tag);
tx_desc++;
i++;
......@@ -1585,31 +1610,21 @@ static void i40e_tx_map(struct i40e_ring *tx_ring, struct sk_buff *skb,
i = 0;
}
frag++;
}
tx_desc->cmd_type_offset_bsz |=
cpu_to_le64((u64)I40E_TXD_CMD << I40E_TXD_QW1_CMD_SHIFT);
i++;
if (i == tx_ring->count)
i = 0;
size = skb_frag_size(frag);
data_len -= size;
tx_ring->next_to_use = i;
dma = skb_frag_dma_map(tx_ring->dev, frag, 0, size,
DMA_TO_DEVICE);
if (tx_flags & (I40E_TX_FLAGS_TSO | I40E_TX_FLAGS_FSO))
gso_segs = skb_shinfo(skb)->gso_segs;
else
gso_segs = 1;
tx_bi = &tx_ring->tx_bi[i];
}
/* multiply data chunks by size of headers */
tx_bi->bytecount = paylen + (gso_segs * hdr_len);
tx_bi->gso_segs = gso_segs;
tx_bi->skb = skb;
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);
/* set the timestamp and next to watch values */
/* set the timestamp */
first->time_stamp = jiffies;
first->next_to_watch = tx_desc;
/* Force memory writes to complete before letting h/w
* know there are new descriptors to fetch. (Only
......@@ -1618,16 +1633,27 @@ static void i40e_tx_map(struct i40e_ring *tx_ring, struct sk_buff *skb,
*/
wmb();
/* set next_to_watch value indicating a packet is present */
first->next_to_watch = tx_desc;
i++;
if (i == tx_ring->count)
i = 0;
tx_ring->next_to_use = i;
/* notify HW of packet */
writel(i, tx_ring->tail);
return;
dma_error:
dev_info(dev, "TX DMA map failed\n");
dev_info(tx_ring->dev, "TX DMA map failed\n");
/* clear dma mappings for failed tx_bi map */
for (;;) {
tx_bi = &tx_ring->tx_bi[i];
i40e_unmap_tx_resource(tx_ring, tx_bi);
i40e_unmap_and_free_tx_resource(tx_ring, tx_bi);
if (tx_bi == first)
break;
if (i == 0)
......@@ -1635,8 +1661,6 @@ static void i40e_tx_map(struct i40e_ring *tx_ring, struct sk_buff *skb,
i--;
}
dev_kfree_skb_any(skb);
tx_ring->next_to_use = i;
}
......
......@@ -103,7 +103,6 @@
#define I40E_TX_FLAGS_FCCRC (u32)(1 << 6)
#define I40E_TX_FLAGS_FSO (u32)(1 << 7)
#define I40E_TX_FLAGS_TXSW (u32)(1 << 8)
#define I40E_TX_FLAGS_MAPPED_AS_PAGE (u32)(1 << 9)
#define I40E_TX_FLAGS_VLAN_MASK 0xffff0000
#define I40E_TX_FLAGS_VLAN_PRIO_MASK 0xe0000000
#define I40E_TX_FLAGS_VLAN_PRIO_SHIFT 29
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册