[PATCH 2/7] [I/OAT] Only offload copies for TCP when there will be a context switch
From: Chris Leech <hidden>
Date: 2006-08-16 00:47:14
Also in:
lkml
Subsystem:
networking [general], networking [tcp], the rest · Maintainers:
"David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Neal Cardwell, Linus Torvalds
The performance wins come with having the DMA copy engine doing the copies in parallel with the context switch. If there is enough data ready on the socket at recv time just use a regular copy. Signed-off-by: Chris Leech <redacted> --- net/ipv4/tcp.c | 10 +++++++--- 1 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 934396b..36f6b64 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c@@ -1105,6 +1105,8 @@ int tcp_recvmsg(struct kiocb *iocb, stru long timeo; struct task_struct *user_recv = NULL; int copied_early = 0; + int available = 0; + struct sk_buff *skb; lock_sock(sk);
@@ -1131,7 +1133,11 @@ int tcp_recvmsg(struct kiocb *iocb, stru #ifdef CONFIG_NET_DMA tp->ucopy.dma_chan = NULL; preempt_disable(); - if ((len > sysctl_tcp_dma_copybreak) && !(flags & MSG_PEEK) && + skb = skb_peek_tail(&sk->sk_receive_queue); + if (skb) + available = TCP_SKB_CB(skb)->seq + skb->len - (*seq); + if ((available < target) && + (len > sysctl_tcp_dma_copybreak) && !(flags & MSG_PEEK) && !sysctl_tcp_low_latency && __get_cpu_var(softnet_data).net_dma) { preempt_enable_no_resched(); tp->ucopy.pinned_list = dma_pin_iovec_pages(msg->msg_iov, len);
@@ -1140,7 +1146,6 @@ int tcp_recvmsg(struct kiocb *iocb, stru #endif do { - struct sk_buff *skb; u32 offset; /* Are we at urgent data? Stop if we have read anything or have SIGURG pending. */
@@ -1428,7 +1433,6 @@ skip_copy: #ifdef CONFIG_NET_DMA if (tp->ucopy.dma_chan) { - struct sk_buff *skb; dma_cookie_t done, used; dma_async_memcpy_issue_pending(tp->ucopy.dma_chan);