Multiple Web Browsers Denial of Service Exploit (1 bug to rule them all)

2009-07-15 / 2009-07-16
Credit: nobody
Risk: Medium
Local: No
Remote: Yes

________________________________________________________________________ One bug to rule them all IE5,IE6,IE7,IE8,Netscape,Firefox,Safari,Opera,Konqueror, Seamonkey,Wii,PS3,iPhone,iPod,Nokia,Siemens.... and more. Don't wet your pants - it's DoS only ________________________________________________________________________ Release mode: Tried hard to coordinate - gave up Reference : [GSEC-TZO-26-2009] - One bug to rule them all WWW : Vendors : others.. Status : Varies CVE : CVE-2009-1692 (created by apple same root cause) Credit : Except Apple - nobody Affected products : ~~~~~~~~~~~~~~~~~ - Internet Explorer 5, 6, 7, 8 (all versions) - Chrome (limited) - Opera - Seamonkey - Midbrowser - Netscape 6 & 8 (9 years ago) - Konqueror (all versions) - Apple iPhone + iPod - Apple Safari - Thunderbird - Nokia Phones : Nokia N95 (Symbian OS v.9.2),Nokia N82, Nokia N810 Internet Tablet - Aigo P8860 (Browser hangs and cannot be restarted) - Siemens phones - Google T-Mobile G1 TC4-RC30 - Ubuntu (Operating system sometimes reboots, memory management failure) - possibly more devices and products that support Javascript, try it yourselves. POC here : Patch availability : ~~~~~~~~~~~~~~~~~~ - Mozilla : Fixed in Firefox 3.0.5 and - Apple iPhone&iPod : patched - IE : No patch for IE5, IE6, IE7, IE8 until IE9 - Webkit : Patched in r41741 - - Chrome : Patched, unknown which version) - Opera : Patched after version 9.64 - Thunderbird (unknown) - Konqueror : unknown (did not respond) - Nokia : unknown, opened a case but never came back - Aigo P8860 : unknown - Siemens : unknown - Others ? Find out by visiting the POC at I. Background ~~~~~~~~~~~ Quoting Wikipedia "ECMAScript is a scripting language, standardized by Ecma International in the ECMA-262 specification and ISO/IEC 16262. The language is widely used on the web, especially in the form of its three best-known dialects, JavaScript, ActionScript, and JScript." II. Description ~~~~~~~~~~~~~ Calling the select() method with a large integer, results in continuos allocation of x+n bytes of memory exhausting memory after a while. The impact varies from null pointer dereference (no more memory,hence crashing the browser) to the reboot of the complete Operation System (Konqueror&Ubuntu) I here had never been a limit specified as to how many html elements the select call should handle, after the report of this Bug, vendors apparently agreed to a limit of 10.000 elements : "Talked to some Apple and Opera guys at the WHATWG social, and we decided this was a good number" III. Impact ~~~~~~~~~ The Impact varies from Browser to Browser and from OS to OS. Here is a small excerpt: - Konqueror (Ubuntu)- allocates 2GB of memory then either crashes the Browser or (most often) the OS reboots. Ubuntu's memory management system appears to be configured as to NOT stop the process that consumes too much memory, but a random process. This sometimes leads to processes that are vital for the OS to be killed, hence the reboot. I am not kidding. Thanks to 'FX' for Memory management hint. - Chrome : allocates 2GB of memory then crashes tab with a null pointer - Firefox : allocates 2GB of memory then the Browser crashes - IE5,6,7,8 : allocates 2GB of memory then the Browser crashes - Opera : Allocated and commits as much memory as available, will not crash but other applications will become unstable - Nintento WII (Opera) : Console hangs, needs hard reset Video: (Thanks to David Raison) - Sony PS3 - Console hangs, needs hard reset Video: (Thanks to Chris Gates) - iPhone - iPhone hangs and needs hard reset Video: (Thanks to g0tcha) - Aigo P8860 (Browser hangs and cannot be restarted) IV. Proof of concept ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ <script> function poc(o) { e = document.createElement("select"); e.length=2147483647; } function go() { poc(0); } </script> URL: Some have not understood what this code does, it does NOT loop as some vendors claimed, it just calls select.lenght() ONCE with a huge integer. One might wonder if over the 9 last years that this bug existed, nobody ever entered a large number in a select.lenght() call. IV. Disclosure timeline ~~~~~~~~~~~~~~~~~~~~~~~ Nothing particular to note, except the usual discussion about availability being a security issue. V. Thanks ~~~~~~~~~~~~~~~~~~~~~~~ Chris Gates, David Raison, Fahem Adam, a team of engineers that recognise themselves and oCert for not helping coordinate this bug.



Vote for this issue:


Thanks for you vote!


Thanks for you comment!
Your message is in quarantine 48 hours.

Comment it here.

(*) - required fields.  
{{ x.nick }} | Date: {{ x.ux * 1000 | date:'yyyy-MM-dd' }} {{ x.ux * 1000 | date:'HH:mm' }} CET+1
{{ x.comment }}

Copyright 2019,


Back to Top