Full Disclosure mailing list archives

Re: Linux (in)security (Was: Re: Re: No Subject)


From: Paul Schmehl <pauls () utdallas edu>
Date: Wed, 22 Oct 2003 22:18:37 -0500

--On Wednesday, October 22, 2003 6:00 PM -0600 Bruce Ediger <eballen1 () qwest net> wrote:

The real questions go something like:

"Source code for Unix viruses has been available for years, from sources
almost too numerous to mention.  Why haven't Unix viruses become epidemic
the way that Windows viruses have?"

The usual argument is that Windows is more ubiquitous than Unix and is therefore the target of choice. I would argue that the *real* reason is that Windows is more ubiquitous as a *desktop* operating system and is therefore the target of choice. However, that's changing. Linux is gaining in the desktop space and so is Mac OS X, which is really "exposed" for the first time. By that I mean that previous Mac OSes weren't as easily attacked remotely because they used Appletalk rather than TCP/IP. (Yes, Macophiles, I know TCP/IP was available before OS X.)

The real key to prevalence of malware, IMNSHO, is the ease of attack *and* the potential pool of victims. People think it's really stupid to "surf" the Internet using an administrator account on Windows. Well what do you think the neophyte Linux users are doing? I seriously doubt you'll find many that have a regular account and use su or sudo to do administrative tasks. They're bound to run in to something sooner or later that they find irritating (like being prompted for root's password every time they try to run up2date on RedHat) and they'll do the same thing they always do on a desktop system. They'll start logging in as root because they don't get "pestered" by all those warning messages and they can install software any time they want. (Mind you, Windows still has a long way to go in that regard. MS doesn't make it easy to run as an unprivileged user, that's for sure.)

And when folks are on the net, logged in as root, on a Unix box, they're just as susceptible to worms and viruses as any Windows user is. All it takes is some momentum in the desktop space and the stats will change. When the average desktop user can figure out how to burn CDs, listen to music and print on *nix as easily as they can do it on Windows, you'll see more and more malware for *nix as they move over to it (if they do.)

Now I am *not* arguing that Windows is the best OS to use (or even a good one for that matter) or even that Windows is no easier to attack than *nix. But worms and viruses will follow desktop users, not OSes, no question about it.

"Security problems of the same magnitude as .ida buffer overflows, or
MSRPC buffer overflows exist in unix programs like Sendmail and others.
Why hasn't a worm materialized for this problem?"

Because unpatched apache isn't installed *and* running on *nix boxes by default. We had 90 boxes hit by Code Red. Only one was an "IT" box, and that one had just been installed and was *at* windowsupdate when it got infected. Of the other 89, all but three were desktop systems. When Nimda hit, we had 40. All 40 were desktops. People who know what they're doing don't get infected with that crap. People who don't, do. What OS they're using is irrelevant.

"The scalper worm didn't effect nearly as many hosts as msblast did.
Why not?  Why did the scalper worm seem to die out, yet wormwatch.org
still records many hits from much older worms like SQLSpida and Nimda?"

Because desktop users don't patch. Scalper didn't make much headway because *very few* desktop *nix boxes run Apache, and servers that do are admined by people who understand the need to patch.

Remember the SunOS.Poisonbox.worm? That made pretty good headway on Solaris boxes and can still be found today. What did it attack? Sadmind, which few server admins would ever run and far fewer would run unpatched. Only desktop users have that on and don't want to be bothered with patching. And they got infected. Every *nix infection that I've had to deal with has been a desktop system, not a server.

Why do you think wuftpd is so heavily attacked? I think it's because it's had many holes *and* lots of desktop users run it because it lets them easily move files around.

And I guess you can generalize and ask why the Windows "culture" generates
so many problems of such a magnitude, that last so long?  My home office
web server got a Code Red hit on Sept 19th 2003, for example.  Other
computing cultures (Unix, Mac, etc) don't seem to exhibit this.  Why not?

Well, historically *nix was for the clued in. All others were excluded. And Mac wasn't easily exploited due to Appletalk. But all that's changing.

KDE has been riddled with security problems. Once the number of desktop users using KDE reaches critical mass (whatever that is) you'll start seeing more and more malware on *nix. Malware follows negligent users, *not* OSes.

Shouldn't we focus our efforts on figuring out what aspects of Linux or
Mac cultures keep epidemics from occuring?  It's certainly a waste of
breath to point out that OS X has horrendous security flaws when none of
them turn into grotesque epidemics like Sobig.f.

Well, think about it for a minute. You're going to write a virus that's designed to trojan machines so you can use them in a massive distributed spam network. What do you attack? The 5 million Mac machines worldwide? Or the 150 million Windows boxes? If your rate of success is 1 in 500, you get 2,000 bots with Mac and 300,000 with Windows. Which would you choose?

I don't doubt that there is some politicization in malware production (people who hate Gates and his OS and want to embarrass him any way they can), but most malware authors are simply trying to get the most bang for the buck, if you will. They'll follow the desktop crowd wherever it leads them. And they won't have any more difficulty infecting KDE users than they do Windows users.

Paul Schmehl (pauls () utdallas edu)
Adjunct Information Security Officer
The University of Texas at Dallas
AVIEN Founding Member
http://www.utdallas.edu

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.netsys.com/full-disclosure-charter.html


Current thread: