There has been a surprising series of kernel security problems reported
over the last week. These include:
- The uselib() vulnerability disclosed
by Paul Starzetz. A locking mistake in an old and mostly unused
system call creates a race condition which can be exploited to change
protections on memory - and compromise the system. The exploit has
not been released, but Mr. Starzetz claims that the race is relatively
easy to exploit by first consuming large amounts of memory to force
the kernel to sleep in the right spot.
- Paul Starzetz also discovered a race
condition in the page fault handler which can only be exploited on
SMP systems. If two threads tried to expand the same downward-growing
memory segment at the same time, the result could be an exploitable
corruption of the page tables.
- The grsecurity team, frustrated at a seeming lack of interest in
security problems among the kernel developers, disclosed five vulnerabilities at once.
One of these is a denial-of-service problem where users could lock
more than the authorized amount of memory into physical RAM; as it
turns out, the kernel developers still are
not overly concerned about that problem. The other
vulnerabilities require root access (or at least access to physical
devices) to exploit; one of them is in a driver which does not compile
in 2.6.
Fixes for the first two vulnerabilities have been merged into the
pre-2.6.11 BitKeeper repository; the last set will be fixed as well, but
with less urgency. Fixes can also be found in the -ac tree and in the updated kernels being
issued by distributors.
One concern that has been raised by these disclosures is that the new
kernel development model, by encouraging such large changes between
releases, is allowing the creation of more security problems. While that
worry could yet prove to be justified, all of the vulnerabilities listed
above, with the exception of the RLIMIT_MEMLOCK denial of service
problem, are present in the 2.4 kernel as well. They were not introduced
or enabled by the new development model.
Another concern is more valid, however: the kernel development project does
not have an official security contact or process for handling security
problems. Developers who know how the kernel process works have no trouble
getting consideration for security-related problems and patches, but the
whole process looks far more opaque to the rest of the world. There is a
clear need for an easily-found contact for kernel security issues. Chris
Wright, who has done a fair amount of security-related kernel work, is pushing for improvements in this area, and,
most importantly, has volunteered to do much of the work. So chances are
this problem will not last much longer.
Comments (11 posted)
New vulnerabilities
bmv: insecure temporary file
| Package(s): | bmv |
CVE #(s): | CAN-2003-0014
|
| Created: | January 11, 2005 |
Updated: | January 12, 2005 |
| Description: |
Peter Samuelson, upstream maintainer of bmv, a PostScript viewer for
SVGAlib, discovered that temporary files are created in an insecure
fashion. A malicious local user could cause arbitrary files to be
overwritten by a symlink attack. |
| Alerts: |
|
Comments (none posted)
dillo: format string vulnerability
| Package(s): | dillo |
CVE #(s): | CAN-2005-0012
|
| Created: | January 10, 2005 |
Updated: | January 12, 2005 |
| Description: |
Gentoo Linux developer Tavis Ormandy found a format string bug in Dillo's
handling of messages in a_Interface_msg(). An attacker could craft a
malicious web page which, when accessed using Dillo, would trigger the
format string vulnerability and potentially execute arbitrary code with the
rights of the user running Dillo. |
| Alerts: |
|
Comments (none posted)
exim: buffer overflows
Comments (1 posted)
hylafax: weak hostname and username validation
| Package(s): | hylafax |
CVE #(s): | CAN-2004-1182
|
| Created: | January 11, 2005 |
Updated: | January 13, 2005 |
| Description: |
Patrice Fournier discovered a vulnerability in the authorization
subsystem of hylafax, a flexible client/server fax system. A local or
remote user guessing the contents of the hosts.hfaxd database could
gain unauthorized access to the fax system. Fixed in HylaFAX
4.2.1. |
| Alerts: |
|
Comments (none posted)
kdelibs: unsanitzied input
| Package(s): | kdelibs |
CVE #(s): | CAN-2004-1165
|
| Created: | January 10, 2005 |
Updated: | July 19, 2005 |
| Description: |
Thiago Macieira discovered a vulnerability in the kioslave library,
which is part of kdelibs, which allows a remote attacker to execute
arbitrary FTP commands via an ftp:// URL that contains an URL-encoded
newline before the FTP command. |
| Alerts: |
|
Comments (none posted)
kernel: race condition, privilege escalation
| Package(s): | kernel |
CVE #(s): | CAN-2004-1235
CAN-2004-1337
|
| Created: | January 10, 2005 |
Updated: | January 19, 2005 |
| Description: |
Paul Starzetz discovered a race condition in the ELF library and a.out
binary format loaders, which can be locally exploited in several
different ways to gain root privileges. (CAN-2004-1235)
Liang Bin found a design flaw in the capability module. After this
module was loaded on demand in a running system, all unprivileged user
space processes got all kernel capabilities (thus essentially root
privileges). (CAN-2004-1337) |
| Alerts: |
|
Comments (none posted)
Konqueror: Java sandbox vulnerabilities
| Package(s): | konqueror |
CVE #(s): | CAN-2004-1145
|
| Created: | January 11, 2005 |
Updated: | January 12, 2005 |
| Description: |
According to this KDE
Security Advisory, two flaws in the Konqueror web browser make it
possible to by pass the sandbox environment which is used to run
Java-applets. All versions of KDE up to KDE 3.3.1 inclusive are affected.
KDE 3.3.2 is not affected. |
| Alerts: |
|
Comments (none posted)
lintian: insecure temporary directory
| Package(s): | lintian |
CVE #(s): | CAN-2004-1000
|
| Created: | January 10, 2005 |
Updated: | January 12, 2005 |
| Description: |
Jeroen van Wolffelaar discovered a problem in lintian, the Debian
package checker. The program removes the working directory even if it
wasn't created at program start, removing an unrelated file or
directory a malicious user inserted via a symlink attack. |
| Alerts: |
|
Comments (none posted)
mailman: cross-site scripting
| Package(s): | mailman |
CVE #(s): | CAN-2004-1177
|
| Created: | January 10, 2005 |
Updated: | March 22, 2005 |
| Description: |
Florian Weimer discovered a cross-site scripting vulnerability in
mailman's automatically generated error messages. An attacker could
craft an URL containing JavaScript (or other content embedded into
HTML) which triggered a mailman error page. When an unsuspecting user
followed this URL, the malicious content was copied unmodified to the
error page and executed in the context of this page. |
| Alerts: |
|
Comments (none posted)
namazu2: cross-site scripting vulnerability
| Package(s): | namazu2 |
CVE #(s): | CAN-2004-1318
|
| Created: | January 6, 2005 |
Updated: | January 12, 2005 |
| Description: |
The namazu2 full text search engine has a cross-site scripting vulnerability
that may allow an attacker to display arbitrarily crafted text
by the use of specially crafted input information. |
| Alerts: |
|
Comments (none posted)
nfs-utils: arbitrary code execution
| Package(s): | nfs-utils |
CVE #(s): | CAN-2004-0946
|
| Created: | January 11, 2005 |
Updated: | February 27, 2006 |
| Description: |
Arjan van de Ven discovered a buffer overflow in rquotad on 64bit
architectures; an improper integer conversion could lead to a buffer
overflow. An attacker with access to an NFS share could send a specially
crafted request which could then lead to the execution of arbitrary code. |
| Alerts: |
|
Comments (none posted)
o3read: buffer overflow during file conversion
| Package(s): | o3read |
CVE #(s): | CAN-2004-1288
|
| Created: | January 11, 2005 |
Updated: | January 12, 2005 |
| Description: |
Wiktor Kopec discovered that
the parse_html function in o3read.c copies any number of bytes into a
1024-byte array. |
| Alerts: |
|
Comments (none posted)
phpgroupware: information disclosure vulnerability
| Package(s): | phpgroupware |
CVE #(s): | |
| Created: | January 6, 2005 |
Updated: | January 12, 2005 |
| Description: |
phpgroupware has multiple vulnerabilities that may
be exploited for the purpose of information disclosure
or a remote compromise. |
| Alerts: |
|
Comments (none posted)
poppassd_pam: unauthorized password changing
| Package(s): | poppassd_pam |
CVE #(s): | CAN-2005-0002
|
| Created: | January 11, 2005 |
Updated: | January 12, 2005 |
| Description: |
Gentoo Linux developer Marcus Hanwell discovered that poppassd_pam did
not check that the old password was valid before changing passwords.
Subsequent investigation revealed that poppassd_pam did not call
pam_authenticate before calling pam_chauthtok. |
| Alerts: |
|
Comments (none posted)
TikiWiki: arbitrary command execution
| Package(s): | TikiWiki |
CVE #(s): | |
| Created: | January 10, 2005 |
Updated: | January 31, 2005 |
| Description: |
TikiWiki lacks a check on uploaded images in the Wiki edit page. A
malicious user could run arbitrary commands on the server by uploading and
calling a PHP script. |
| Alerts: |
|
Comments (none posted)
UnRTF: Buffer overflow
| Package(s): | unrtf |
CVE #(s): | |
| Created: | January 11, 2005 |
Updated: | January 12, 2005 |
| Description: |
An unchecked strcat() in unrtf may overflow the bounds of a static buffer.
Using a specially crafted file, possibly delivered by e-mail or over the
web, an attacker may execute arbitrary code with the permissions of the
user running UnRTF. |
| Alerts: |
|
Comments (1 posted)
vilistextum: buffer overflow vulnerability
| Package(s): | vilistextum |
CVE #(s): | CAN-2004-1299
|
| Created: | January 6, 2005 |
Updated: | January 12, 2005 |
| Description: |
Vilistextum has a buffer overflow vulnerability that can
allows an attacker
to execute arbitrary code via a maliciously created web page. |
| Alerts: |
|
Comments (none posted)
Resources
Version v2.3 of the Metasploit Framework is out. "
The 2.3 release includes three user interfaces,
46 exploits and 68 payloads.
"
Full Story (comments: none)
Page editor: Jonathan Corbet
Next page:
Kernel development>>