From: Eric Dumazet <edumazet@google.com>
To: "David S . Miller" <davem@davemloft.net>
Cc: netdev <netdev@vger.kernel.org>,
Willem de Bruijn <willemb@google.com>,
Eric Dumazet <edumazet@google.com>,
Eric Dumazet <eric.dumazet@gmail.com>
Subject: [PATCH v2 net-next 1/2] net: prepare (struct ubuf_info)->refcnt conversion
Date: Thu, 31 Aug 2017 16:48:21 -0700 [thread overview]
Message-ID: <20170831234822.26612-2-edumazet@google.com> (raw)
In-Reply-To: <20170831234822.26612-1-edumazet@google.com>
In order to convert this atomic_t refcnt to refcount_t,
we need to init the refcount to one to not trigger
a 0 -> 1 transition.
This also removes one atomic operation in fast path.
v2: removed dead code in sock_zerocopy_put_abort()
as suggested by Willem.
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
net/core/skbuff.c | 10 ++--------
net/ipv4/tcp.c | 2 --
2 files changed, 2 insertions(+), 10 deletions(-)
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 917da73d3ab3b82163cf0a9ee944da09cb5a391f..1a754d7896ceac1eb85e3b13c1422b4d6a88fedf 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -963,7 +963,7 @@ struct ubuf_info *sock_zerocopy_alloc(struct sock *sk, size_t size)
uarg->len = 1;
uarg->bytelen = size;
uarg->zerocopy = 1;
- atomic_set(&uarg->refcnt, 0);
+ atomic_set(&uarg->refcnt, 1);
sock_hold(sk);
return uarg;
@@ -1005,6 +1005,7 @@ struct ubuf_info *sock_zerocopy_realloc(struct sock *sk, size_t size,
uarg->len++;
uarg->bytelen = bytelen;
atomic_set(&sk->sk_zckey, ++next);
+ sock_zerocopy_get(uarg);
return uarg;
}
}
@@ -1102,13 +1103,6 @@ void sock_zerocopy_put_abort(struct ubuf_info *uarg)
atomic_dec(&sk->sk_zckey);
uarg->len--;
- /* sock_zerocopy_put expects a ref. Most sockets take one per
- * skb, which is zero on abort. tcp_sendmsg holds one extra, to
- * avoid an skb send inside the main loop triggering uarg free.
- */
- if (sk->sk_type != SOCK_STREAM)
- atomic_inc(&uarg->refcnt);
-
sock_zerocopy_put(uarg);
}
}
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 21ca2df274c5130a13d31a391a1408d779af34af..fff4d7bc3b8b46174e7bd0a04d7c212307e55cb5 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1190,8 +1190,6 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
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;
}
--
2.14.1.581.gf28d330327-goog
next prev parent reply other threads:[~2017-08-31 23:48 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-31 23:48 [PATCH v2 net-next 0/2] net: ubuf_info.refcnt conversion Eric Dumazet
2017-08-31 23:48 ` Eric Dumazet [this message]
2017-08-31 23:54 ` [PATCH v2 net-next 1/2] net: prepare (struct ubuf_info)->refcnt conversion Willem de Bruijn
2017-08-31 23:48 ` [PATCH v2 net-next 2/2] net: convert (struct ubuf_info)->refcnt to refcount_t Eric Dumazet
2017-08-31 23:55 ` Willem de Bruijn
2017-09-01 0:04 ` [PATCH v2 net-next 0/2] net: ubuf_info.refcnt conversion Eric Dumazet
2017-09-01 17:36 ` Eric Dumazet
2017-09-02 3:22 ` David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170831234822.26612-2-edumazet@google.com \
--to=edumazet@google.com \
--cc=davem@davemloft.net \
--cc=eric.dumazet@gmail.com \
--cc=netdev@vger.kernel.org \
--cc=willemb@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).