NetBSD Problem Report #47055
From dholland@netbsd.org Wed Oct 10 19:44:55 2012
Return-Path: <dholland@netbsd.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
by www.NetBSD.org (Postfix) with ESMTP id F39DC63B90D
for <gnats-bugs@gnats.NetBSD.org>; Wed, 10 Oct 2012 19:44:54 +0000 (UTC)
Message-Id: <20121010194454.E495214A155@mail.netbsd.org>
Date: Wed, 10 Oct 2012 19:44:54 +0000 (UTC)
From: dholland@NetBSD.org
Reply-To: dholland@NetBSD.org
To: gnats-bugs@gnats.NetBSD.org
Subject: libc diagassert issues
X-Send-Pr-Version: 3.95
>Number: 47055
>Category: lib
>Synopsis: libc diagassert issues
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: lib-bug-people
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Oct 10 19:45:00 +0000 2012
>Closed-Date: Sun Oct 06 12:56:52 +0000 2013
>Last-Modified: Sun Oct 06 12:56:52 +0000 2013
>Originator: David A. Holland
>Release: NetBSD 6.99.12 (20121010)
>Organization:
>Environment:
n/a
>Description:
There are two problems with libc diagassert:
(1) it calls syslog() blindly, so if an assertion happens in a program
that hasn't called openlog(), it ends up using uninitialized state in
syslog.c. The observable effect of this seems to be to issue syslog
messages with no hostname field; however, it could be (or become,
without much notice) a good deal worse.
(2) it isn't thread-safe. It frob a global variable (diagassert_flags)
without locking. It also should probably be using syslog_r() rather
than syslog().
>How-To-Repeat:
Build kde4, look at the resulting syslog spam. Then review the code.
>Fix:
Probably needs discussion.
>Release-Note:
>Audit-Trail:
From: christos@zoulas.com (Christos Zoulas)
To: gnats-bugs@NetBSD.org, lib-bug-people@netbsd.org,
gnats-admin@netbsd.org, netbsd-bugs@netbsd.org
Cc:
Subject: Re: lib/47055: libc diagassert issues
Date: Wed, 10 Oct 2012 18:45:29 -0400
On Oct 10, 7:45pm, dholland@NetBSD.org (dholland@NetBSD.org) wrote:
-- Subject: lib/47055: libc diagassert issues
| There are two problems with libc diagassert:
|
| (1) it calls syslog() blindly, so if an assertion happens in a program
| that hasn't called openlog(), it ends up using uninitialized state in
| syslog.c. The observable effect of this seems to be to issue syslog
| messages with no hostname field; however, it could be (or become,
| without much notice) a good deal worse.
This is a bug. I am fixing and versioning syslog(). Syslog should be
thread-safe, and because of the static hostname (and the initialiazation
of it, it is not).
christos
From: "Christos Zoulas" <christos@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/47055 CVS commit: src/lib/libc/gen
Date: Wed, 10 Oct 2012 18:50:51 -0400
Module Name: src
Committed By: christos
Date: Wed Oct 10 22:50:51 UTC 2012
Modified Files:
src/lib/libc/gen: syslog.c
Log Message:
PR/47055: David Holland: Remove static hostname and store the hostname in
syslog_data. Initialize hostname before it is used. While here, add a version
to syslog_data.
To generate a diff of this commit:
cvs rdiff -u -r1.50 -r1.51 src/lib/libc/gen/syslog.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: "Christos Zoulas" <christos@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/47055 CVS commit: src/sys/sys
Date: Wed, 10 Oct 2012 18:51:55 -0400
Module Name: src
Committed By: christos
Date: Wed Oct 10 22:51:55 UTC 2012
Modified Files:
src/sys/sys: syslog.h
Log Message:
PR/47055: David Holland: Add hostname to syslog_data, version number, rename
fields consistently, and version library calls affected.
To generate a diff of this commit:
cvs rdiff -u -r1.34 -r1.35 src/sys/sys/syslog.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: matthew green <mrg@eterna.com.au>
To: gnats-bugs@NetBSD.org
Cc: lib-bug-people@netbsd.org, gnats-admin@netbsd.org,
netbsd-bugs@netbsd.org
Subject: re: lib/47055: libc diagassert issues
Date: Thu, 11 Oct 2012 16:53:50 +1100
> (1) it calls syslog() blindly, so if an assertion happens in a program
> that hasn't called openlog(), it ends up using uninitialized state in
> syslog.c. The observable effect of this seems to be to issue syslog
> messages with no hostname field; however, it could be (or become,
> without much notice) a good deal worse.
that shouldn't be a problem. it sounds like a bug in syslog(3) itself.
i've never seen any requirement for openlog(3) being necessary and indeed
our documentation says:
The openlog() function provides for more specialized processing of the
messages sent by syslog() and vsyslog().
> (2) it isn't thread-safe. It frob a global variable (diagassert_flags)
> without locking. It also should probably be using syslog_r() rather
> than syslog().
probably. (i don't think we had syslog_r() back when diagassert
first appeared...)
.mrg.
State-Changed-From-To: open->closed
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Sun, 06 Oct 2013 12:56:52 +0000
State-Changed-Why:
fixed last year
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.