NetBSD Problem Report #53775

From he@smistad.uninett.no  Tue Dec 11 09:43:58 2018
Return-Path: <he@smistad.uninett.no>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(Client CN "mail.NetBSD.org", Issuer "mail.NetBSD.org CA" (not verified))
	by mollari.NetBSD.org (Postfix) with ESMTPS id 92B9B7A1C5
	for <gnats-bugs@gnats.NetBSD.org>; Tue, 11 Dec 2018 09:43:58 +0000 (UTC)
Message-Id: <20181211094351.8D9C843F3D1@smistad.uninett.no>
Date: Tue, 11 Dec 2018 10:43:51 +0100 (CET)
From: he@NetBSD.org
Reply-To: he@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: bind(2) may inaccurately return EADDRNOTAVAIL
X-Send-Pr-Version: 3.95

>Number:         53775
>Category:       kern
>Synopsis:       bind(2) may inaccurately return EADDRNOTAVAIL
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Dec 11 09:45:00 +0000 2018
>Last-Modified:  Tue Dec 11 14:40:00 +0000 2018
>Originator:     he@NetBSD.org
>Release:        NetBSD 8.0_STABLE
>Organization:
	I try...
>Environment:
System: NetBSD smistad.uninett.no 8.0_STABLE NetBSD 8.0_STABLE (GENERIC) #0: Sun Aug 5 00:07:14 CEST 2018 he@smistad.uninett.no:/usr/obj/sys/arch/amd64/compile/GENERIC amd64
Architecture: x86_64
Machine: amd64
>Description:
	The bind(2) system call may return EADDRNOTAVAIL also in cases
	where it more precisely should return EADDRINUSE.

	This was uncovered as part of an investigation as to why BIND
	in some rare cases returns SERVFAIL for no good reason, but
	that problem I have not come to the bottom of.

>How-To-Repeat:
	Inspect the code.

	Compare to FreeBSD, and a change they did way back in 2007
	(which doesn't mention the changed error codes in the log
	entry, and also does other stuff) at

https://svnweb.freebsd.org/base/head/sys/netinet/sctp_pcb.c?r1=170090&r2=170091&

>Fix:
--- sys/netinet/sctp_pcb.c.orig 2018-09-03 18:29:36.000000000 +0200
+++ sys/netinet/sctp_pcb.c      2018-12-11 10:17:10.000000000 +0100
@@ -1751,41 +1751,41 @@ sctp_inpcb_bind(struct socket *so, struc
                SCTP_INP_WUNLOCK(inp);
                inp_tmp = sctp_pcb_findep(addr, 0, 1);
                if (inp_tmp != NULL) {
                        /* lock guy returned and lower count
                         * note that we are not bound so inp_tmp
                         * should NEVER be inp. And it is this
                         * inp (inp_tmp) that gets the reference
                         * bump, so we must lower it.
                         */
                        SCTP_INP_WLOCK(inp_tmp);
                        SCTP_INP_DECR_REF(inp_tmp);
                        SCTP_INP_WUNLOCK(inp_tmp);

                        /* unlock info */
                        SCTP_INP_INFO_WUNLOCK();
-                       return (EADDRNOTAVAIL);
+                       return (EADDRINUSE);
                }
                SCTP_INP_WLOCK(inp);
                if (bindall) {
                        /* verify that no lport is not used by a singleton */
                        if (sctp_isport_inuse(inp, lport)) {
                                /* Sorry someone already has this one bound */
                                SCTP_INP_DECR_REF(inp);
                                SCTP_INP_WUNLOCK(inp);
                                SCTP_INP_INFO_WUNLOCK();
-                               return (EADDRNOTAVAIL);
+                               return (EADDRINUSE);
                        }
                }
        } else {
                /*
                 * get any port but lets make sure no one has any address
                 * with this port bound
                 */

                /*
                 * setup the inp to the top (I could use the union but this
                 * is just as easy
                 */
                uint32_t port_guess;
                uint16_t port_attempt;
                int not_done=1;

>Audit-Trail:
From: "Christos Zoulas" <christos@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/53775 CVS commit: src/sys/netinet
Date: Tue, 11 Dec 2018 09:38:46 -0500

 Module Name:	src
 Committed By:	christos
 Date:		Tue Dec 11 14:38:46 UTC 2018

 Modified Files:
 	src/sys/netinet: sctp_pcb.c

 Log Message:
 PR/53775: Havard Eidnes: bind(2) may inaccurately return EADDRNOTAVAIL,
 it should return EADDRINUSE.


 To generate a diff of this commit:
 cvs rdiff -u -r1.17 -r1.18 src/sys/netinet/sctp_pcb.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

Home	PR Database Search