linux-kernel - [PATCH] sctp: if backlog is 0, listening shall not be deactivated.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 14 Jan 2009 10:42:03 +0100
From:	Pierre Habouzit <pierre.habouzit@...ersec.com>
To:	Wei Yongjun <yjwei@...fujitsu.com>,
	Vlad Yasevich <vladislav.yasevich@...com>,
	" David S. Miller" <davem@...emloft.net>
Cc:	linux-kernel@...r.kernel.org,
	Pierre Habouzit <pierre.habouzit@...ersec.com>
Subject: [PATCH] sctp: if backlog is 0, listening shall not be deactivated.

POSIX hints that when 0 is used for the listen backlog argument, the
kernel shall chose a default automatic value. TCP for example, works this
way.

Signed-off-by: Pierre Habouzit <pierre.habouzit@...ersec.com>
---

 net/sctp/socket.c |   20 --------------------
 1 files changed, 0 insertions(+), 20 deletions(-)

    To put a bit of background, I stumbled against this while doing a code
    that basically did:

	struct sctp_event_subscribe events;
	/* ... */

	fd = socket(AF_INET, SOCK_SEQPACKETS, IPPROTO_SCTP);
	sctp_bindx(fd, ....);
	events = (struct sctp_event_subscribe){
	    .sctp_data_io_event     = 1,
	    .sctp_association_event = 1,
	};
	setsockopt(fd, SOL_SCTP, SCTP_EVENTS, &events, sizeof(events));
	listen(fd, 0);
	len = sctp_recvmsg(fd, .....);

    The latter call instead of blocking like I expected returned with
    errno == ENOTCONN.

    I know POSIX doesn't _require_ listen() to accept 0 as a valid backlog,
    but the other listen() implementation I have used in the kernel do, and
    it looks really surprising for the programmer (who really searches the
    error elsewhere).

    Fortunately I had another working code at hand and I managed to find the
    problem resorting to reverting the changes I made to the original code
    line per line (*sigh*).

    I'm unsure if the diff shouldn't do instead:

    if (!backlog)
        backlog = 1;

    I'm not really comfortable around the kernel core ;)


diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index ff0a8f8..da1d96a 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -5866,16 +5866,6 @@ SCTP_STATIC int sctp_seqpacket_listen(struct sock *sk, int backlog)
 	if (!sctp_style(sk, UDP))
 		return -EINVAL;
 
-	/* If backlog is zero, disable listening. */
-	if (!backlog) {
-		if (sctp_sstate(sk, CLOSED))
-			return 0;
-
-		sctp_unhash_endpoint(ep);
-		sk->sk_state = SCTP_SS_CLOSED;
-		return 0;
-	}
-
 	/* Return if we are already listening. */
 	if (sctp_sstate(sk, LISTENING))
 		return 0;
@@ -5919,16 +5909,6 @@ SCTP_STATIC int sctp_stream_listen(struct sock *sk, int backlog)
 	struct sctp_sock *sp = sctp_sk(sk);
 	struct sctp_endpoint *ep = sp->ep;
 
-	/* If backlog is zero, disable listening. */
-	if (!backlog) {
-		if (sctp_sstate(sk, CLOSED))
-			return 0;
-
-		sctp_unhash_endpoint(ep);
-		sk->sk_state = SCTP_SS_CLOSED;
-		return 0;
-	}
-
 	if (sctp_sstate(sk, LISTENING))
 		return 0;
 
-- 
1.6.1.161.g5e07b.dirty

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/