Slashdot Mirror
Zero Day Hole In Google Desktop
40by40 writes "A Web application security specialist has figured out a way to launch man-in-the-middle attacks against a computer with a fully patched Google Desktop installed. With knowledge of the Google Desktop security model (a combination of one-time tokens, iFrames and JavaScript), hacker Robert Hansen figured out a way to sit between a target launching a Google search query and manipulate the search results to take control of other programs on the desktop. From the article: 'This should drive home the point that deep integration between the desktop and the web is not a good idea, without tremendous thought put into the security model. As Google's site is unencrypted, and they place their content that can run executables on their site, it can be subverted by an attacker," Hansen warns. Hansen's advisory comes just days after a Chris Soghoian's exposé of a similar man-in-the-middle attack scenario against a remote vulnerability in the upgrade mechanism used by a number of commercial Firefox extensions.'"
Google should stop screwing around and just bite the bullet: develop your own operating system based on Linux and get it over with. Windows Vista is down, kick them in the nuts when you can!
Some say he is made with ascii, others that he is eyeballed daily by millions. All we know is, he is known as the Sig
This should drive home the point that connections should flow over encrypted tunnels whenever possible, to reduce the ease of performing man in the middle attacks. If this session flowed over an SSL style connection, the man in the middle would first need to figure out how to get into that session. That strategy seriously reduces the places where malicious code can exist "in the middle". Don't throw the baby (rich client interaction with services in the cloud) out with the bathwater.
If you mod me down, I shall become more powerful than you could possibly imagine.
By now, everybody developing browser components should know that you do not provide functions which can execute arbitrary programs.
Usually, it's Microsoft doing this, with Outlook, IE, Office, etc. launching other applications. This is the source of most of the vulnerabilities involving web browsing. Now we have Google competing to offer similar security holes.
Google apps are getting more popular--> Ditto
When the popularity of a software approach a critical mass, the probability of exploits appearing approachs 1.
It's time to realise that Abble's products are the biggest abomination these days. Just say NO to the dumb iAbble way!!
Will the google fanboys please step up and tell us how this doesn't matter?
It sounds like this takes advantage of the "Google Integration" feature, where the Google Desktop software adds a link to your Google search results page. I found his explanation rather unclear, but it sounds like you can avoid this by going into Google Desktop's preferences, then the Display tab, then un-checking the last checkbox, "Show Desktop Search results on Google Web Search result pages".
I've always thought that was a scary idea anyway, since my desktop content should be in a clearly-partitioned security domain from Web content.
Once you are compromised this way the attack tries to take advantage of cross scripting vulnerabilities in a browser to run code in the compromised machine. I am not sure if there is anything unique to Google Desktop here. Could the same attack take advantage of the numerous ActiveX vulnerabilities?
Is the "security expert" trying to get more mileage by listing each exploitable hole of a man-in-the-middle attack as a separate discovery?
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
How does one stop Google desktop from indexing executables? When I open the Google Desktop preferences, exe files aren't even listed as something I can index, but search for an executable like hypertrm.exe on Google desktop, and it shows up anyway, which is the 'meat' of this vulnerability.
It should be illegal to say that freedom of speech should be limited.
Anyone want to bet that this is the beginning of a little landslide?
I wish the Google team all the best in dealing with this issue... but I am scratching my head at the speed with which they are attempting to diversify their offerings.
Google did not become a dominant force overnight. They fought battles, learned lessons, and refined/defined search capabilities for the entire world. Why have they been shooting off in a dozen different directions? Is there any way that even they can stay on top of all the little details considering the number of immature products they are floating?
Anyhow, the next couple of days will go a long way towards showing exactly how far the Google team needs to go before I trust them on my desktop. Here's hoping they prove to have the response time/customer centric attitude that made them my preferred search provider.
Regards.
"Tremendous thought" is a weaker notion than transparency, public scrutiny, or even rigorous proof, which are really what's required.
Everything else is just hope; hide and seek.
Hopefully Google can learn and set an example here.
There are no karma whores, only moderation johns
this is even more of a problem since more and more installers like Irfanview's or Adobe's include Google Desktop (and/or toolbar) and there is no way to skip them when doing automated installs... what a sick trend.
It's time for this app-in-the-browser stupidity to stop. The web browser was initially developed to display documents, and allow for navigation between such documents. Since then, we've had a variety of parties add all sorts of crap onto that: JavaScript, ActiveX, Java applets, Flash, AJAX, etc. And what have we found? A lot of trouble, and little benefit.
As is shown here, security holes become a very major problem. The software itself becomes difficult to develop, let alone test on the wide variety of browsers that are out there. And even then, the software itself doesn't perform very well, and lacks many features.
We already have a variety of secure, sensible methods of remotely viewing desktops. X11 over SSH is one option. VNC is another. So perhaps what Google should be doing is offering their hosted services, but instead of going through a web browser, they should provide a simple, easy-to-use X11 server for Windows that works well over low-powered connections. Those of us running Linux, BSD or Mac OS X can directly use the X11 capabilities that are usually inherent to our OSes. Regardless, we connect to Google's servers, have our data stored there, and just view it remotely. Hell, we could even run a web browser (Firefox, Opera, Konqueror, etc.) on their system if we really needed to view web pages from within their client application.
The more shit we throw onto a browser, the crappier it'll get. We'll be facing more and more security glitches, browser bloat, and shitty experiences.
This guy is probably funded by M$. I mean, come on. Hello, Mr. FUD. You want to see dangerous deep integration? Internet Explorer. Durr. I have a news flash for this genius. Pretty much nothing is a good idea without giving careful consideration to security. Things like: installing software on your computer (any software), clicking on links in a browser, typing text into your computer, saving files to disk, taking a dump. Yep, pretty much all of them are potentially dangerous.
Web-desktop integration is already here and it isn't going anywhere. It's a perfectly good idea, not a bad one. And because it's a good idea and because it involves your data, it's also a good idea to address security concerns. That is the fair and unbiased statement.
Did the industry and Google learn nothing from the mistakes Microsoft made?
Even MS has done a 180 and with Vista broke all the internal/external links that made XP/ActiveX/IE such a mess. So if MS is smart enough to learn from their mistakes you would thing a company like Google would not go out of their way to emulate the same bad security ideas.
Is it just me, or is Google racing to be the next big evil? Gmail scanning, search data compiling, Firefox reporting, desktop document reporting, and now making really stupid software design decisions?
I think the premise of the article is rather stupid in fact.
It is not Google's job to provide a secure channel.
I guess when I do a MITM attack to capture login prompts and transparently proxy that is google's problem also?
Or when I resolve DNS queries to my own box, that is likewise google at fault?
Lol.
It's the phrase which springs to mind with "web 2.0" applications. You have an exposed API on both sides, the client and the server.
Deleted
Plan 9?
installing third-party applications that connect to someplace, download something, and do something on on your machine, and being exposed when those applications are shown to have bugs is news how?
;)
the google engineers aren't magicians. when they develop features, they do so under tight schedule, and make mistakes, especially those hired to code (as opposed to do PR). the only reason there haven't been more problems discovered is likely the fact that they don't distribute much software.
besides, google's main goal isn't promoting security. their primary goal is to hookup lotsa people -- and in their case, that means to deliver applications with lotsa features quickly, because people are hooked on the features, the competition ain't sleeping, and that first-comer advantage matters.
does that remind you of another company? it should, because all of them successful companies ain't that much different at all
Google is nice enough to offer SSL for most of its services these days. It would make a lot of sense for them to round out their secure offerings with an SSL search as well.
Right now, any request to an encrypted Google search URL redirects you to www.google.com.
I seriously doubt it. Goatse's hole is really big.
Hrm... you seem unaware that the very desktop (and mobile) friendly Macintosh and the coming generation of iPhones, iPods, and probably other digital appliances from Apple are based on a real UNIX underneath? The UNIX foundation of the system design is partly responsible for the rapid pace of evolution of Mac OS X.
Although extreme hubris might combine with extreme resources (both dollars and talent) at Google to lead to the creation of an entirely new OS from the ground up, there may not be any need for that. The UNIX wheel is relatively round these days, particularly considering the Mac OS X / OSX example. Better yet, UNIX is nicely modular. If anyone devises a clever way to "avoid buffer overflow situations" it seems likely, on the basis of past evidence concerning technology development and adoption within UNIX systems in general, that it would be easier to integrate that language and compiler, or whatever technology it happens to be, into a UNIX operating system than it would be to create a fully capable system on top of it from whole cloth.
Since you seem genuinely interested in the topic, here are some reasonable books on operating system design which you might enjoy.
The Design and Implementation of the 4.4 BSD Operating System
Design of the UNIX Operating System
Operating System Design: The Xinu Approach
UNIX Internals: The New Frontiers
Mac OS X Internals: A Systems Approach
Solaris Internals
The other issues you raise are largely issues of interface design, which the open source community seems to do rather poorly, or at least not as well as it does other things. Google certainly does not need to re-invent the entire operating system wheel to improve URL integration, or provide a "minimalist" desktop interface, for example. They don't even need to strip features, really. Mac OS X, for example, provides enough of a minimalist default interface that novice computer users are comfortable with it. A Linux based OS from Google could take a similar approach, perhaps being even more spartan in the basic features, if that's really a desirable goal (which is another question entirely).
If you mod me down, I shall become more powerful than you could possibly imagine.
We'd better get used to Google becoming the butt of jokes usually aimed at ActiveX. Google Gears, Google Desktop, Google whatever. We now reaize that the developers that develop these technologies simply get traded between the big 3 (Google, MS, Yahoo) and others.
Are we all finally realizing that Google writes insecure apps just like ever other software development company that is made up of humans?
Have you seen gears.google.com? It's technology that lets you run web apps offline. Basically, this gets rid of one of the biggest complaints about web apps like google spreadsheet and docs. I think technology like this will eventually put the nail in the coffin for a large class of desktop-only apps.
Just how prevalent are these men who are in the middle? I've yet to hear about an actual attacker using this strategy. Is that because the middle men are pretty much undetectable and many compromises happen without the user noticing that he didn't do anything 'wrong?' The crackers seem to have an easy enough time phishing their way into your data or doing social engineering to land an executable on your machine. It seems like it's much harder to set yourself up as this man in the middle than it is to find exploits or engineer your way in. No amount of SSL will save users from sending data to the bad guys by impersonation. Once a cracker finds his exploit and has landed code or a file in an arbitrary location on your machine, you're compromised, and the gig is up for the user. Even SSL is vulnerable, since it is so inconvenient, nay, near impossible to run many programs as a limited user under XP. Maybe Vista makes progress, maybe the user clicks so often that he gets duped into nullifying this "security."
That said, I think a browser actually accessing non-"browser system" files on your disk without a warning of some kind is a bad idea. Clicking on links is exploit sensitive, sure, but why make being a man in the middle more than a data collecting trick?
Today is all we really have. We should all live it well: it is our stepping stone to all of our tomorrows.
is a disconnected computer.
This mad rush to put everything online is, well, mad..
Robert Hansen
Major U.S. software companies should really consider nuking Scandinavia?
You want to see sandboxing? IE on Vista. Durr.
Now you are forced to run JS to get all the "news" features, and running a search takes you to a blank query screen instead of parsing your web search terms into automagically a news search when you mash the "news" button.
Seriously borked and lame. They call it an improvement, I call it they just doubled the complexity of a simple search for no apparent reason other than they could and have a boner now for "everything JS" (or even worse, Flash) like so many other web developer dweebs
Hell, it's working good but we have to justify our jobs! Let's type some crap up and make it more stupid and complicated! yaaaa! Pass another can of red bull!
PhDs = piled higher and deeper sometimes. Or walk !chewgum
Why on this Earth would Google want an OS?? They already have it - it is called "The Browser". That's what they use to make money. They may want to extend its usage, but I doubt that Google will ever want to deal with the "desktop" in the same way as Microsoft, Apple or Linux community.
Google is about control. They want to control your information for their own profit. They show it again and again. That's how they make money. The more targeted the ads, the more money they can make. The only competitor I think they may have here is Amazon, but that only deals with your book preferences. Google wants your wants so they can sell something from one of their customers.
Thus it is NOT in the interest of Google to make a desktop. They are not in the business of making software like MS or Apple or GNU or even IBM. They are in business to manage information about you and me. Their "free" solutions are just there so you can give them more info about yourself.
Hope that is clear enough.
... as a pebble disturbs the stillness of the pond? [Ti Kwan Leep]
A programmer who is too proud to think about how other people solved the problem they're looking at is much more likely to invent a wheel with some number of road-contact surfaces "n" where n > 1.
UNIX has survived (indeed thrives) as a result of a number of major refactoring efforts, directed not only at improving the internal architecture, but even the underlying abstractions. Consider Mach and the microkernel revolution, which resulted in nearly every major operating system kernel being refactored to accomodate the design abstractions described in Programming Under Mach. And now class, let us rejoin the mind to the body and gaze into the heart of the candle in meditation.
If you mod me down, I shall become more powerful than you could possibly imagine.
a windows based google OS. Whatttttt? stop looking at me like that.
Well, I'm sorry to be the bearer of bad news, but you haven't spelled anything out. In fact, you've accidentally helped me develop my case. We'll get to that in a moment, but first let me mention that interlocking design elements of the CPU, compilers, and programing languages combine to make buffer overflows possible.
To the extent that portions of POSIX are specified in terms of C or assume C language features, and to the extent that such dependencies upon the nature of the C language, compiler, and host CPU make buffer overflows possible, then, yes, you could say POSIX is broken. However, there is a corpse missing. If it were POSIX that were responsible for buffer overflows, then Microsoft's nortoriously broken implementation of POSIX on Windows NT/2000/XP should have either insulated them from buffer overflows (because it didn't work any better for worm authors than for anybody else), or exposed them to myriad POSIX exploiting buffer overflows (because it somewhat worked, and the presence of POSIX was responsible for buffer overflows). In point of painfully obvious fact, neither happened. It is the Win32 and related Microsoft API, not POSIX, that exposed Windows users to rather more buffer overflows than all the factually POSIX compliant systems on the planet combined. Yeah, the POSIX API is a little long in the tooth, and sure, parts of it are sub-optimal, broken if you like, but certainly not directly and soley responsible for buffer overflows as you imply. Good grief that's a silly notion. Where are all the POSIX worms? Last I checked, Win32 and application worms dominated.
Furthermore, it's certainly possible to dramatically reduce the exposure of a UNIX operating system like Linux or Mac OS X to buffer overflows, by re-implementing certain widely used network-facing services in a more secure language like Java. Why are BIND, Sendmail, IMAP servers, file servers like Samba, and application servers like Apache still largely written in C based languages? POSIX certainly isn't to blame for that. You can't even blame the limited POSIX security model, since it's been extended with more modern ACLs for quite some time. In most cases, we can't even blame language performance issues. In contrast with its reputation for poor performance on GUI tasks, Java gets pretty good marks for raw server side type benchmarks. Whey don't we see one or more of these projects refactoring in Java? Nothing about the modular flexible nature of the UNIX architecture prevents this.
The FUD about Mach is likewise painfully tedious. Mach's infamous performance problems apply to research versions of the kernel that neither NeXT, nor Apple, nor DEC, nor IBM, nor anybody else who's using Mach parts in their kernel ever shipped. Aside from being utterly irrelevant on the grounds that no shipping product was observed to have them, the performance problems were largely due to issues that don't apply when the UNIX server is compiled in with the Mach kernel, as it is with the BSD server in Mac OS X. Those infamous performance problems were due to the research attempt to run the API server in user space, so that the kernel could support multiple "OS personalities", and so that an execution thread could be made more easily migratable from one host OS instance to another. Those were a couple of interesting goals that appealed to a few geeks and super computer researchers, but were not relevant to a desktop or server operating system. Nothing much was sacrificed when Mach and BSD were united in one happy binary.
Conduct a little thought experiment. If there existed a massive performance delta in raw kernel performance between, say, Mac OS X and Linux, it would be easy to demonstrate, if, say, you could run both kernels on the same hardware easily, say, an iMac, Mac Mini, MacBook, etc. These problems would be widely documented by now, many months after the Intel platform equalized the hardware b
If you mod me down, I shall become more powerful than you could possibly imagine.
Ah, the old fallacy that "secure" is a binary value, i.e. that something is either secure or not, and the false conclusion that all apps that aren't 100% secure are all equally insecure.
Comment removed based on user account deletion
Tossing stuff out to users with security holes is something that has earned Microsoft a reputation they'd rather not have. And this kind of bad Microsoft practice is certainly something Google would not want to emulate. So Google had better nip this in the bud quickly, especially as they continue to roll out new products at a rapid pace.
I will happily eat crow if Google ever produces a Linux desktop, but gut instinct says that they won't. So don't get your hopes up.
I hope you aren't a vegetarian...
Google has already developed their own operating system based on Debian, but it's strictly in-house for the time being.