A format string flaw was found in the Python CGI Kit (neo_cgi) module of ClearSilver, a language-neutral HTML templating system, processed certain input, leading to Common Gateway Interface (CGI) script errors. A remote attacker could provide a specially-crafted input, which once processed by an application, using the Python language API of ClearSilver neo_cgi module, could lead to that particular application crash, or, potentially arbitrary code execution with the privileges of the user running the application. References: [1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=649322 Patch, proposed by the issue reporter to the Debian Bug Tracking System: [2] http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=5;filename=fix-cgi-error-format-security.patch;att=1;bug=649322
This issue affects the versions of the clearsilver package, as shipped with Fedora EPEL 4, 5, and 6 releases. Please schedule an update. -- This issue affects the versions of the clearsilver package, as shipped with Fedora release of 14, 15, and 16. Please schedule an update.
Created clearsilver tracking bugs for this issue Affects: fedora-all [bug 757543] Affects: epel-all [bug 757544]
CVE request: [3] http://www.openwall.com/lists/oss-security/2011/11/27/1
Created attachment 537196 [details] Local copy of patch, proposed in the Debian Bug Tracking System
Follow up from Colin Watson on this ( [4] http://www.openwall.com/lists/oss-security/2011/11/27/2 ): On Sun, Nov 27, 2011 at 06:21:15PM +0100, Jan Lieskovsky wrote: > a format string flaw was found in the Python CGI Kit (neo_cgi) > module of ClearSilver, a language-neutral HTML templating system, > processed certain input, leading to Common Gateway Interface (CGI) > script errors. A remote attacker could provide a specially-crafted > input, which once processed by an application, using the Python > language API of ClearSilver neo_cgi module, could lead to that > particular application crash, or, potentially arbitrary code > execution with the privileges of the user running the application. Thanks for responding to this. FWIW, I've attached a copy of the original mail I sent to a couple of security@ addresses about this vulnerability. -- Colin Watson [cjwatson] Subject: clearsilver: possible format string vulnerability in Python extension From: Colin Watson <cjwatson> Date: 11/17/2011 06:12 PM To: security, security CC: clearsilver.org While doing the Perl 5.14 transition in Ubuntu, I noticed that clearsilver has a -Wformat-security warning (Ubuntu builds with -Werror=format-security by default to catch exactly this kind of problem): gcc -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -g -O2 -fstack-protector --param=ssp-buffer-size=4 -Wformat -Wformat-security -g -O2 -fstack-protector --param=ssp-buffer-size=4 -Wformat -Wformat-security -Werror=format-security -Wall -fPIC -Wall -I.. -D_FORTIFY_SOURCE=2 -fPIC -I/usr/include/python2.7 -I.. -D_FORTIFY_SOURCE=2 -fPIC -I../ -I/usr/include/python2.7 -c neo_cgi.c -o build/temp.linux-i686-2.7/neo_cgi.o neo_cgi.c: In function 'p_cgi_error': neo_cgi.c:181:3: error: format not a string literal and no format arguments [-Werror=format-security] The effects of this can be reproduced like this: $ python >>> import neo_cgi >>> cgi = neo_cgi.CGI() >>> cgi.error('%s') Status: 500 Content-Type: text/html <html><body> An error occured:<pre>|▒U▒LfU▒LfU▒@▒` ▒`▒y▒, x▒</pre></body></html> In fact, the examples shipped with clearsilver include exception handlers that call cgi.error(s), so if you can manage to get a % into something that will end up in a Python traceback then you can read bits of process memory over the Internet and possibly do a limited amount of modification too (with %n). I have not reported this upstream. http://code.google.com/p/clearsilver/source/browse/trunk/python/neo_cgi.c shows that it has not yet been fixed. Upstream appears to be http://www.clearsilver.net/ / blong; perhaps somebody could coordinate with him if you confirm this as a possible vulnerability? Thanks, -- Colin Watson [cjwatson]
This was assigned CVE-2011-4357: http://www.openwall.com/lists/oss-security/2011/11/28/6