mirror of https://gitee.com/openkylin/linux.git
net/mlx4_en: Saving mem access on data path
Localized the pdev->dev, and using dma_map instead of pci_map There are multiple map/unmap operations on data path, optimizing those by saving redundant pointer access. Those places were identified as hot-spots when running kernel profiling during some benchmarks. The fixes had most impact when testing packet rate with small packets, reducing several % from CPU load, and in some case being the difference between reaching wire speed or being CPU bound. Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il> Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
parent
6975f4ce5a
commit
ebf8c9aa03
|
@ -1062,6 +1062,7 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
|
|||
memset(priv, 0, sizeof(struct mlx4_en_priv));
|
||||
priv->dev = dev;
|
||||
priv->mdev = mdev;
|
||||
priv->ddev = &mdev->pdev->dev;
|
||||
priv->prof = prof;
|
||||
priv->port = port;
|
||||
priv->port_up = false;
|
||||
|
|
|
@ -48,7 +48,6 @@ static int mlx4_en_alloc_frag(struct mlx4_en_priv *priv,
|
|||
struct mlx4_en_rx_alloc *ring_alloc,
|
||||
int i)
|
||||
{
|
||||
struct mlx4_en_dev *mdev = priv->mdev;
|
||||
struct mlx4_en_frag_info *frag_info = &priv->frag_info[i];
|
||||
struct mlx4_en_rx_alloc *page_alloc = &ring_alloc[i];
|
||||
struct page *page;
|
||||
|
@ -72,7 +71,7 @@ static int mlx4_en_alloc_frag(struct mlx4_en_priv *priv,
|
|||
skb_frags[i].offset = page_alloc->offset;
|
||||
page_alloc->offset += frag_info->frag_stride;
|
||||
}
|
||||
dma = pci_map_single(mdev->pdev, page_address(skb_frags[i].page) +
|
||||
dma = dma_map_single(priv->ddev, page_address(skb_frags[i].page) +
|
||||
skb_frags[i].offset, frag_info->frag_size,
|
||||
PCI_DMA_FROMDEVICE);
|
||||
rx_desc->data[i].addr = cpu_to_be64(dma);
|
||||
|
@ -186,7 +185,6 @@ static void mlx4_en_free_rx_desc(struct mlx4_en_priv *priv,
|
|||
struct mlx4_en_rx_ring *ring,
|
||||
int index)
|
||||
{
|
||||
struct mlx4_en_dev *mdev = priv->mdev;
|
||||
struct page_frag *skb_frags;
|
||||
struct mlx4_en_rx_desc *rx_desc = ring->buf + (index << ring->log_stride);
|
||||
dma_addr_t dma;
|
||||
|
@ -198,7 +196,7 @@ static void mlx4_en_free_rx_desc(struct mlx4_en_priv *priv,
|
|||
dma = be64_to_cpu(rx_desc->data[nr].addr);
|
||||
|
||||
en_dbg(DRV, priv, "Unmapping buffer at dma:0x%llx\n", (u64) dma);
|
||||
pci_unmap_single(mdev->pdev, dma, skb_frags[nr].size,
|
||||
dma_unmap_single(priv->ddev, dma, skb_frags[nr].size,
|
||||
PCI_DMA_FROMDEVICE);
|
||||
put_page(skb_frags[nr].page);
|
||||
}
|
||||
|
@ -412,7 +410,6 @@ static int mlx4_en_complete_rx_desc(struct mlx4_en_priv *priv,
|
|||
int length)
|
||||
{
|
||||
struct skb_frag_struct *skb_frags_rx = skb_shinfo(skb)->frags;
|
||||
struct mlx4_en_dev *mdev = priv->mdev;
|
||||
struct mlx4_en_frag_info *frag_info;
|
||||
int nr;
|
||||
dma_addr_t dma;
|
||||
|
@ -435,7 +432,7 @@ static int mlx4_en_complete_rx_desc(struct mlx4_en_priv *priv,
|
|||
goto fail;
|
||||
|
||||
/* Unmap buffer */
|
||||
pci_unmap_single(mdev->pdev, dma, skb_frag_size(&skb_frags_rx[nr]),
|
||||
dma_unmap_single(priv->ddev, dma, skb_frag_size(&skb_frags_rx[nr]),
|
||||
PCI_DMA_FROMDEVICE);
|
||||
}
|
||||
/* Adjust size of last fragment to match actual length */
|
||||
|
@ -461,7 +458,6 @@ static struct sk_buff *mlx4_en_rx_skb(struct mlx4_en_priv *priv,
|
|||
struct mlx4_en_rx_alloc *page_alloc,
|
||||
unsigned int length)
|
||||
{
|
||||
struct mlx4_en_dev *mdev = priv->mdev;
|
||||
struct sk_buff *skb;
|
||||
void *va;
|
||||
int used_frags;
|
||||
|
@ -483,10 +479,10 @@ static struct sk_buff *mlx4_en_rx_skb(struct mlx4_en_priv *priv,
|
|||
/* We are copying all relevant data to the skb - temporarily
|
||||
* synch buffers for the copy */
|
||||
dma = be64_to_cpu(rx_desc->data[0].addr);
|
||||
dma_sync_single_for_cpu(&mdev->pdev->dev, dma, length,
|
||||
dma_sync_single_for_cpu(priv->ddev, dma, length,
|
||||
DMA_FROM_DEVICE);
|
||||
skb_copy_to_linear_data(skb, va, length);
|
||||
dma_sync_single_for_device(&mdev->pdev->dev, dma, length,
|
||||
dma_sync_single_for_device(priv->ddev, dma, length,
|
||||
DMA_FROM_DEVICE);
|
||||
skb->tail += length;
|
||||
} else {
|
||||
|
|
|
@ -198,7 +198,6 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
|
|||
struct mlx4_en_tx_ring *ring,
|
||||
int index, u8 owner)
|
||||
{
|
||||
struct mlx4_en_dev *mdev = priv->mdev;
|
||||
struct mlx4_en_tx_info *tx_info = &ring->tx_info[index];
|
||||
struct mlx4_en_tx_desc *tx_desc = ring->buf + index * TXBB_SIZE;
|
||||
struct mlx4_wqe_data_seg *data = (void *) tx_desc + tx_info->data_offset;
|
||||
|
@ -214,7 +213,7 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
|
|||
if (likely((void *) tx_desc + tx_info->nr_txbb * TXBB_SIZE <= end)) {
|
||||
if (!tx_info->inl) {
|
||||
if (tx_info->linear) {
|
||||
pci_unmap_single(mdev->pdev,
|
||||
dma_unmap_single(priv->ddev,
|
||||
(dma_addr_t) be64_to_cpu(data->addr),
|
||||
be32_to_cpu(data->byte_count),
|
||||
PCI_DMA_TODEVICE);
|
||||
|
@ -223,7 +222,7 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
|
|||
|
||||
for (i = 0; i < frags; i++) {
|
||||
frag = &skb_shinfo(skb)->frags[i];
|
||||
pci_unmap_page(mdev->pdev,
|
||||
dma_unmap_page(priv->ddev,
|
||||
(dma_addr_t) be64_to_cpu(data[i].addr),
|
||||
skb_frag_size(frag), PCI_DMA_TODEVICE);
|
||||
}
|
||||
|
@ -241,7 +240,7 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
|
|||
}
|
||||
|
||||
if (tx_info->linear) {
|
||||
pci_unmap_single(mdev->pdev,
|
||||
dma_unmap_single(priv->ddev,
|
||||
(dma_addr_t) be64_to_cpu(data->addr),
|
||||
be32_to_cpu(data->byte_count),
|
||||
PCI_DMA_TODEVICE);
|
||||
|
@ -253,7 +252,7 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
|
|||
if ((void *) data >= end)
|
||||
data = ring->buf;
|
||||
frag = &skb_shinfo(skb)->frags[i];
|
||||
pci_unmap_page(mdev->pdev,
|
||||
dma_unmap_page(priv->ddev,
|
||||
(dma_addr_t) be64_to_cpu(data->addr),
|
||||
skb_frag_size(frag), PCI_DMA_TODEVICE);
|
||||
++data;
|
||||
|
@ -733,7 +732,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
|
|||
/* Map fragments */
|
||||
for (i = skb_shinfo(skb)->nr_frags - 1; i >= 0; i--) {
|
||||
frag = &skb_shinfo(skb)->frags[i];
|
||||
dma = skb_frag_dma_map(&mdev->dev->pdev->dev, frag,
|
||||
dma = skb_frag_dma_map(priv->ddev, frag,
|
||||
0, skb_frag_size(frag),
|
||||
DMA_TO_DEVICE);
|
||||
data->addr = cpu_to_be64(dma);
|
||||
|
@ -745,7 +744,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
|
|||
|
||||
/* Map linear part */
|
||||
if (tx_info->linear) {
|
||||
dma = pci_map_single(mdev->dev->pdev, skb->data + lso_header_size,
|
||||
dma = dma_map_single(priv->ddev, skb->data + lso_header_size,
|
||||
skb_headlen(skb) - lso_header_size, PCI_DMA_TODEVICE);
|
||||
data->addr = cpu_to_be64(dma);
|
||||
data->lkey = cpu_to_be32(mdev->mr.key);
|
||||
|
|
|
@ -482,6 +482,7 @@ struct mlx4_en_priv {
|
|||
struct mlx4_en_stat_out_mbox hw_stats;
|
||||
int vids[128];
|
||||
bool wol;
|
||||
struct device *ddev;
|
||||
};
|
||||
|
||||
enum mlx4_en_wol {
|
||||
|
|
Loading…
Reference in New Issue