tcp: enable MSG_ZEROCOPY

Enable support for MSG_ZEROCOPY to the TCP stack. TSO and GSO are
both supported. Only data sent to remote destinations is sent without
copying. Packets looped onto a local destination have their payload
copied to avoid unbounded latency.

Tested:
  A 10x TCP_STREAM between two hosts showed a reduction in netserver
  process cycles by up to 70%, depending on packet size. Systemwide,
  savings are of course much less pronounced, at up to 20% best case.

  msg_zerocopy.sh 4 tcp:

  without zerocopy
    tx=121792 (7600 MB) txc=0 zc=n
    rx=60458 (7600 MB)

  with zerocopy
    tx=286257 (17863 MB) txc=286257 zc=y
    rx=140022 (17863 MB)

  This test opens a pair of sockets over veth, one one calls send with
  64KB and optionally MSG_ZEROCOPY and on the other reads the initial
  bytes. The receiver truncates, so this is strictly an upper bound on
  what is achievable. It is more representative of sending data out of
  a physical NIC (when payload is not touched, either).

Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 9dd6f4dba9b136153054e2c5a3db59c695fc9b0a..71b25567e787f9f93dd0e132a15545ddc6ae3a00 100644 (file)
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1165,6 +1165,7 @@ static int tcp_sendmsg_fastopen(struct sock *sk, struct msghdr *msg,
 int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
 {
        struct tcp_sock *tp = tcp_sk(sk);
+       struct ubuf_info *uarg = NULL;
        struct sk_buff *skb;
        struct sockcm_cookie sockc;
        int flags, err, copied = 0;
@@ -1174,6 +1175,26 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
        long timeo;
 
        flags = msg->msg_flags;
+
+       if (flags & MSG_ZEROCOPY && size) {
+               if (sk->sk_state != TCP_ESTABLISHED) {
+                       err = -EINVAL;
+                       goto out_err;
+               }
+
+               skb = tcp_send_head(sk) ? tcp_write_queue_tail(sk) : NULL;
+               uarg = sock_zerocopy_realloc(sk, size, skb_zcopy(skb));
+               if (!uarg) {
+                       err = -ENOBUFS;
+                       goto out_err;
+               }
+
+               /* skb may be freed in main loop, keep extra ref on uarg */
+               sock_zerocopy_get(uarg);
+               if (!(sk_check_csum_caps(sk) && sk->sk_route_caps & NETIF_F_SG))
+                       uarg->zerocopy = 0;
+       }
+
        if (unlikely(flags & MSG_FASTOPEN || inet_sk(sk)->defer_connect)) {
                err = tcp_sendmsg_fastopen(sk, msg, &copied_syn, size);
                if (err == -EINPROGRESS && copied_syn > 0)
@@ -1297,7 +1318,7 @@ new_segment:
                        err = skb_add_data_nocache(sk, skb, &msg->msg_iter, copy);
                        if (err)
                                goto do_fault;
-               } else {
+               } else if (!uarg || !uarg->zerocopy) {
                        bool merge = true;
                        int i = skb_shinfo(skb)->nr_frags;
                        struct page_frag *pfrag = sk_page_frag(sk);
@@ -1335,6 +1356,13 @@ new_segment:
                                page_ref_inc(pfrag->page);
                        }
                        pfrag->offset += copy;
+               } else {
+                       err = skb_zerocopy_iter_stream(sk, skb, msg, copy, uarg);
+                       if (err == -EMSGSIZE || err == -EEXIST)
+                               goto new_segment;
+                       if (err < 0)
+                               goto do_error;
+                       copy = err;
                }
 
                if (!copied)
@@ -1381,6 +1409,7 @@ out:
                tcp_push(sk, flags, mss_now, tp->nonagle, size_goal);
        }
 out_nopush:
+       sock_zerocopy_put(uarg);
        return copied + copied_syn;
 
 do_fault:
@@ -1397,6 +1426,7 @@ do_error:
        if (copied + copied_syn)
                goto out;
 out_err:
+       sock_zerocopy_put_abort(uarg);
        err = sk_stream_error(sk, flags, err);
        /* make sure we wake any epoll edge trigger waiter */
        if (unlikely(skb_queue_len(&sk->sk_write_queue) == 0 &&