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>] [day] [month] [year] [list]
Message-ID: <a1aaf36e0712291015l1fecd25cyb45beff03b2618a1@mail.gmail.com>
Date: Sat, 29 Dec 2007 19:15:58 +0100
From: "Lolek of TK53" <lolek1337@...glemail.com>
To: vulnwatch@...nwatch.org, full-disclosure@...ts.grok.org.uk,
	bugtraq@...urityfocus.com
Subject: TK53 Advisory #2: Multiple vulnerabilities in ClamAV

========================================================================

                    TK53 Advisory #2 12/29/2007

    - Multiple vulnerabilities in ClamAV

========================================================================

* Authors: Roflek of TK53 <roflek1337@...il.com>, Lolek of TK53
<lolek1337@...il.com>

* Affected program: ClamAV (http://www.clamav.net/)

* Affected versions: 0.92

* Overwiew:
  1) ClamAV uses own functions to create temporary files. One such routine is
  vulnerable to a race condition attack.

  2) ClamAV fails to properly check for base64-UUEncoded files, allowing
     bypassing of the scanner through the use of such files.

  3) The sigtool utility included in the ClamAV distribution fails to handle
     created files in a secure way.


==========================================
== Race Condition Vulnerability Details ==
==========================================

$SOURCE/libclamav/others.c (line 488):

int cli_gentempfd(const char *dir, char **name, int *fd)
{

    *name = cli_gentemp(dir);
    if(!*name)
        return CL_EMEM;

    *fd = open(*name, O_RDWR|O_CREAT|O_TRUNC|O_BINARY, S_IRWXU);
    if(*fd == -1) {
        cli_errmsg("cli_gentempfd: Can't create temporary file %s:
%s\n", *name, strerror(errno));
        free(*name);
        return CL_EIO;
    }

    return CL_SUCCESS;
}

This function, cli_gentempfd, uses a custom function to generate a
(more or less)
unique file name which is then opened, and the file descriptor is returned via
an output parameter.

The problem with this code is that a race condition exists: if the attacker is
able to guess the generated file name, he/she is able to create such a named
file between the call of cli_gentemp() and open(), making it possible to
overwrite arbitrary files to which the user that runs ClamAV has write access
with temporary data. A solution to fix this problem is to use the O_EXCL option
for open(). This option prevents that the file will be opened if it already
exists.

So, how does the file name generation happen? First, cli_gentemp() determines
the temporary directory. Users of the cli_gentemp() function can specify their
own custom temporary directory. If none is specified, then the content of the
TMPDIR environment variable is used. If the environment variable is unset, then
P_tmpdir resp. "/tmp" are used. The generated format of the file name is
$TMPDIR/clamav-$HASH, where $HASH is generated from a fixed 16 byte "salt" and
32 (more or less) random bytes.

The salt is defined in the following way:

static unsigned char name_salt[16] = { 16, 38, 97, 12, 8, 4, 72, 196,
217, 144, 33, 124, 18, 11, 17, 253 };

The random bytes are generated with an internal function cli_rndnum()
which looks
like this:

unsigned int cli_rndnum(unsigned int max)
{
    struct timeval tv;

  gettimeofday(&tv, (struct timezone *) 0);
  srand(tv.tv_usec+clock());

  return rand() % max;
}

As you can see, every time cli_rndnum() is called, the random number generator
is reinitialized with the microsecond component of the current time and an
"approximation of the processor time used by the program" using the clock()
function. This takes away a lot of randomness from the value returned by
cli_rndnum(): as seed, more or less public information which should be
relatively easy to be guessed by the attacker is used, making it possible to
guess the value returned by rand(). Also, since the random number generator is
reseeded every time cli_rndnum() is called, every returned value is directly
computed from the seed.

In addition, cli_rndnum() uses the modulo operator to "cut off" the random
number at a maximum value, which is discouraged by virtually every documentation
of the rand() function. The publication "Numerical Recipes in C: The Art of
Scientific Computing"[0] says about the use of rand():

	"If you want to generate a random integer between 1 and 10, you
should always do it by using high-order bits, as in

		j=1+(int) (10.0*rand()/(RAND_MAX+1.0));

	and never by anything resembling

		j=1+(rand() % 10);

	(which uses lower-order bits)."


The function cli_gentempfd() is used throughout the whole ClamAV source code in
numerous places, which means that all these places are affected by the race
conditions.  Ironically, the code also uses cli_gentemp() in several places to
generate a random file name and then passes the file name to call to open()
with the O_EXCL option enabled.

The race condition was introduced to the ClamAV source code on August 31st,
2007, in SVN revision 3196. The first release that contains the bug was 0.92.
Since then, the code has remained in the trunk of the SVN repository.


======================================================
== Base64 UUEncoded Files Scanner Bypassing Details ==
======================================================

ClamAV contains functionality to unpack and scan different types of files, such
as archive files. Beside others, UUEncoded files are supported, too.

But ClamAV a variation of the UUEncoded file format: uuencode implementations,
such as the one supplied by the GNU sharutils package, support the use of
BASE64 instead of the uuencode-proprietary encoding, making it possible to
create uuencode files slightly smaller. These files start with "begin-base64"
instead of the usual "begin " of UUEncoded files.

When ClamAV tries to determine a file's type through its magic number at the
beginning, it correctly recognizes files that start with "begin ", but fail to
recognize Base64-UUEncoded files, since it's not in the list of checked magic
numbers.

This makes it possible for attackers to bypass scanning of archive files by
packing e.g. malware into a Base64-UUEncoded file, which the user then opens.

As history has shown, social-engineering-based attacks with unknown file
attachments have been successful a lot of times. This fact makes such an attack
sound more realistic, especially since many users blindly trust their virus
scanner. In cases where ClamAV is the only virus scanner in the mail delivery
process, and then fails to scan Base64-UUEncoded malware, this would be even
more serious, since naive users may open unknown files without suspicion "since
the virus scanner has checked them already and found nothing".

===============================================
== Insecure File Handling in Sigtool Details ==
===============================================

Sigtool as shipped in a ClamAV installation is vulnerable to insecure
file handling
in some cases. If Sigtool is used with the 'utf16-decode' it fails to
check if the
<filename>.ascii file already exists and thus allowing an attacker
with write permissions
to the directory in which <filename> is located (for example if Sigtool is used
in /tmp/) to overwrite arbitrary files belonging to the user via a
symlink attack.

sigtool.c:
156 fname = opt_arg(opt, "utf16-decode");
157 if((fd1 = open(fname, O_RDONLY)) == -1) {
158     mprintf("!utf16decode: Can't open file %s\n", fname);
159     return -1;
160 }
161
162 newname = malloc(strlen(fname) + 7);
163 sprintf(newname, "%s.ascii", fname);
164
165 if((fd2 = open(newname, O_WRONLY|O_CREAT|O_TRUNC, S_IRWXU)) < 0) {
166     mprintf("!utf16decode: Can't create file %s\n", newname);

open is missing O_EXCL and thanks to the usage of O_TRUNC the file will be
truncated to size zero.

== References ==

[0]: Numerical Recipes in C: The Art of Scientific Computing (William H. Press,
	Brian P. Flannery, Saul A. Teukolsky, William T. Vetterling; New York:
	Cambridge University Press, 1992 (2nd ed., p. 277))

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ