RE: RE: [VulnWatch] Rendering large binary file as HTML makes Mozilla Firefox stop responding or crash
"Troy Fletcher" <[email protected]> Tue, 26 Oct 2004 11:46:25 -0700
| Newsgroups | gmane.comp.security.vulnerabilities.watch.general |
|---|---|
| Message-ID | <[email protected]> |
Peter, From what I can tell, I'm running XP Pro SP2, IE 6.0 SP2 (though I might be misreading the version number listed in the about internet explorer box), "Version:6.0.2900.2180.xpsp_sp2_rtm.040803-2158" (simple, right?) "Update Versions:; SP2;" It's not likely I'm running an older version than you are. When I run the binary html file on IE I get the same garbage that is displayed in the Mozilla browser, which should not happen if IE was recognizing it as a binary. Is it possible that some IE setting allows it to make a best guess at the actual file type as opposed to trusting the file extension? I don't think it has anything to do with zones (I ran it locally), because you did the same with the file once it was uploaded to a remote website. BTW: I ran the same test on a 5MB executable (the Firefox setup program :) and IE was able to load the file as an html (a full page of garbage), and increased memory usage about 20%. After it loaded I was able to close or navigate away from the page easily, and memory returned. When I did the same on Firefox memory usage increased about 60%, and took much longer to load completely. Once it was loaded my system was strained, and closing the page took a long time, and took even longer for the memory to return. -Troy -----Original Message----- From: Peter Kruse [mailto:[email protected]] Sent: Tuesday, October 26, 2004 9:51 AM To: Troy Fletcher; [email protected] Subject: SV: [VulnDiscuss] RE: [VulnWatch] Rendering large binary file as HTML makes Mozilla Firefox stop responding or crash Hi Troy, >I renamed my SP2 executable (big enough? :) to .html and opened it >up in firefox while watching page fault rate and cpu, and did the >same with IE. Opening the file in IE was considerably harder on my >machine than with firefox, both did crash though (or at least >didn't respond fast enough for windows). Did you do something I >didn't? I'm sure you tested this with IE, right? The size of that file should be enough to trigger this flaw. However, smaller files will also do the trick ;-) I tested this with no impact on IE 6.0, SP1, running on Windows XP with Servicepack 2. I was prompted if I wanted to run the executable. Regards Peter Kruse >-----Original Message----- >From: Peter Kruse [mailto:[email protected]] >Sent: Monday, October 25, 2004 1:44 AM >To: [email protected] >Subject: [VulnWatch] Rendering large binary file as HTML makes Mozilla >Firefox stop responding or crash > > > >-----BEGIN PGP SIGNED MESSAGE----- >Hash: SHA1 > >CSIS Security Advisory: [CSIS2004-5) > >Rendering large binary file as HTML makes Mozilla Firefox stop >responding or >crash > >Date Published: 10.25.2004 (GMT) > >Summary >======== >Mozilla Firefox, Web-browser built for 2004, advanced e-mail and newsgroup >client, IRC chat client, and HTML editing made simple. The Mozilla Firefox >shippes with several bugs, making it possible to crash the browser, eat up >virtual memory, simply by hosting a binary renamed as html, on a remote >website. > >Vulnerability Class >=================== >The browser should remain responsive while displaying large files. Instead >it crashes and hangs and feeds on virtual memory which could cause the >operating system to become unstable. > >Details >======= >Internet Explorer, and other browsers, verifies the content of filetypes >before opening in the browser. Based on the content of the file, it decides >what application should be used to open/view the content of the file. This >is, by design, not the case with Mozilla based browsers. A >malicious website >can host a large chunck of data, spoofed as a html file that Mozilla will >display within the browser window. Thereby effectively causing a crash on >systems visiting the website. > >You can choose any file from your harddisk larger than 5MB, rename it as a >html file, upload it to a remote website, or simply open it directly from >your local harddrive. The result is the same: Mozilla will stop responding, >showing a lot of binary garbage (clearly understandable), before >the user is >forced to either end the application or reboot the system. > >In several test scenarios the system force feed all virtual memory causing >the system to become unstable. However, this all depends on the size of the >file viewed by the browser. To avoid the user from being suspicious while >the file loads and garbage is showed in the browser window you can format >the website in such a way that garbage won't show. This way the >browser will >show a blank page until it crashes and the system becomes unstable. When >viewed, the browser will load the binary without the users knowledge. The >fact that this bug can be trigged by sending the same file with 1024 ASCII >characters pre-pended makes exploitation trivial. > >Impact >====== >Low-Medium: This is a remote DoS in Mozilla Firefox. There are >several other >ways to crash the browser. > >This behavior was confirmed with Mozilla/5.0 (Windows; U; Windows NT 5.1; >rv:1.7.3) Gecko/20040913 Firefox/0.10, but my guess is that all versions of >Mozilla introduce the problem. > >Solution >========= >Awaiting fix > >Affected Products >================ >Mozilla/5.0 Gecko/20040913 Firefox/0.10 and prior > > >- --- > >Med venlig hilsen // Kind regards > >Peter Kruse, >Security- and virusanalyst, >CSIS ApS >http://www.csis.dk > >PGP fingerprint >79FD 0648 158E 6B9E 236F CFDA 7C58 64D6 BE83 FA60 > >Combined Services & Integrated Solutions >Gevnø Gade 11a >4660 Store Heddinge, Denmark > >-----BEGIN PGP SIGNATURE----- >Version: PGP 8.1 > >iQA/AwUBQXy8J3xYZNa+g/pgEQLy1gCeIOBSUFvWcMDxRdctMJKZyepxBuUAn0cs >2AJ7hwekVBENB2m1+t5CoQ26 >=Mi5B >-----END PGP SIGNATURE----- > >