2014-04-14 17:39:37

by Kenjiro Nakayama

Subject: [PATCH] ipv4: Add option to get TCP_FASTOPEN to getsockopt()

TCP_FASTOPEN option can be set via setsockopt(), but the value cannot be
gotten via getsockopt(). This patch adds the option to getsockopt().

Sighned-off-by: Kenjiro Nakayama <[email protected]>

Add option to get TCP_FASTOPEN to getsockopt()
---
net/ipv4/tcp.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 4bd6d52..6ccdbcc 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2916,6 +2916,11 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
case TCP_USER_TIMEOUT:
val = jiffies_to_msecs(icsk->icsk_user_timeout);
break;
+
+ case TCP_FASTOPEN:
+ val = sysctl_tcp_fastopen;
+ break;
+
case TCP_TIMESTAMP:
val = tcp_time_stamp + tp->tsoffset;
break;
--
1.9.0

2014-04-15 00:43:29

by Neal Cardwell

Subject: Re: [PATCH] ipv4: Add option to get TCP_FASTOPEN to getsockopt()

On Mon, Apr 14, 2014 at 1:39 PM, Kenjiro Nakayama
<[email protected]> wrote:
> TCP_FASTOPEN option can be set via setsockopt(), but the value cannot be
> gotten via getsockopt(). This patch adds the option to getsockopt().
>
> Sighned-off-by: Kenjiro Nakayama <[email protected]>
>
> Add option to get TCP_FASTOPEN to getsockopt()
> ---
> net/ipv4/tcp.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
> index 4bd6d52..6ccdbcc 100644
> --- a/net/ipv4/tcp.c
> +++ b/net/ipv4/tcp.c
> @@ -2916,6 +2916,11 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
> case TCP_USER_TIMEOUT:
> val = jiffies_to_msecs(icsk->icsk_user_timeout);
> break;
> +
> + case TCP_FASTOPEN:
> + val = sysctl_tcp_fastopen;
> + break;
> +
> case TCP_TIMESTAMP:
> val = tcp_time_stamp + tp->tsoffset;
> break;
> --
> 1.9.0

Hi,

Please cc the Linux network development mailing list at
[email protected] (cc-ed) for Linux networking patches like this.

I don't think this is the implementation we'd want for a getsockopt()
for TCP_FASTOPEN. I imagine that since the setsockopt(TCP_FASTOPEN)
enables Fast Open on a listening socket with a given backlog, if we
add a corresponding getsockopt() then it should return that Fast Open
backlog if Fast Open is enabled on the listening socket, or 0
otherwise. That is, basically it would be querying whether
fastopen_init_queue() has been run and returning fastopenq->max_qlen
if so. That way setsockopt(TCP_FASTOPEN, getsockopt(TCP_FASTOPEN))
will be a NOP in most cases, as one would expect.

neal