The Security Nerd

How someone could tap your optical fiber

2007-10-07T16:45:00.000-07:00

With photos of equipment and discussion of how it works:
http://blogs.techrepublic.com.com/security/?p=222&tag=nl.e036

Do you need n-gram frequencies for your crypto work?

2007-09-07T13:34:00.000-07:00

Or just for statistics in some other line of work?

How would you like to get n-gram frequencies from Google's corpus? How would you like to have a training corpus of a trillion words?

Google Research has published them, on 6 DVDs. Announcement at http://googleresearch.blogspot.com/2006/08/all-our-n-gram-are-belong-to-you.html

Technical article about the Greek wiretapping case

2007-07-05T19:42:00.000-07:00

Technical details of the 2004-2005 intercepts of Greek government officials's mobile phone calls. It was much more than just turning on the "lawful intercept" functions: the code turned off log files and used rootkit techniques.

Software key logger countermeasures reviewed

2007-04-26T18:26:00.000-07:00

via Rootsecure, Informatica review of programs designed to block software keyloggers. This isn't about detection and removal, as a typical antimalware package would do. These are programs that encrypt or hide keystreams so that any keylogger that gets installed sees nothing or sees a scrambled stream of keystrokes.

Interestingly, several products limit their scope to protecting keystrokes in web browsers.

Good review of hardware keyloggers

2007-04-26T17:59:00.000-07:00

The Iron Geek reviews PS/2 and USB keystroke loggers.

Since the days of the keystroke recorder that looks like an RF suppressor in the cable, there's been a new generation of USB devices that, with varying and sometimes configurable stealthiness, sit on the USB bus and record keyboard traffic as it goes by.

There's no real defense except for physical security.

Untappable fiber?

2007-04-26T00:36:00.000-07:00

If you know much about the physics of optical fiber, you know that there are ways to make light leak out without breaking the fiber.

That news is now widely known, since The Register has published an article about optical fiber eavesdropping with Exfo's FCD-10B coupler. I'm skeptical about the description of a "simple clip-on" device, given the amount of sheathing and armor on fiber lines, but Infoguard alleges that someone found an eavesdropping device on a Verizon fiber line in 2003.

Who's Infoguard? They sell encryption solutions for high-speed fiber.

Rutkowska strikes again!

2007-03-03T18:09:00.000-08:00

Most forensics work these days involves shutting down a system and studying the hard disk in isolation. The problem is that if you're studying malware and it's on a critical system or is written by somebody clever, it might live entirely in RAM.

So the arms race continued with ways to copy live RAM onto an acquisition device. Which of course any self-respecting rootkit can subvert.

So the arms raced continued with proposals to build hardware RAM-acquisition boards which would go into a PCI slot and use DMA to read system RAM. Foolproof, right?

Unless the malware author is as smart as Joanna Rutkowska, who observed that DMA requests do not go through the same mechanisms as CPU RAM access does, that the difference is configurable, and that malware with driver-level access can

Remap the DMA access from the acquisition card to be memory-mapped IO pointing back to the card, thus crashing the system that you couldn't afford to take down for study
Redirect that memory-mapped IO to another PCI card and "cover" a set of addresses with unrelated bits in place of those the CPU sees
Write to that other PCI card and control what the acquisition card sees.

This is beautiful work. It's theoretically detectable: it would take consummate artistry to change a block of RAM and have the result be internally and logically consistent.

If you are trying to prevent attacks like that, I don't see any way offhand. She studied AMD systems but there's no reason to think that others are less configurable.

Joanna Rutkowska presentation at BlackHat DC on defeating hardware-based rootkit detectors.

Enigma machine animation

2007-02-22T23:42:00.000-08:00

Michael Heyman, on a cryptography mailing list, points to a .

Sandbox IE with Greenborder: pointers to tests

2007-01-26T21:56:00.000-08:00

Ian Richards tested security-through-virtualization products, including deliberately infecting his machine with CoolWebSearch. As of midyear 2006, the best-performing of eight products was Greenborder.
See also a test commissioned by Greenborder.

There were some early and severe compatibility problems, for example with Nod32.

Google fixes fascinating vulnerability

2007-01-16T19:31:00.000-08:00

For a while it was possible to put your own content on a *.google.com domain. This would allow stealing Google cookies, and as Tony Ruscoe points out, it would be hard for even a savvy user to avoid the problem.

Excerpts won't do the article justice. Read the whole thing.

Can you read old data from overwritten flash memory?

2007-01-04T10:07:00.000-08:00

It's actually analog in the sense that hard disks are analog: the physical layer stores a bunch of electrons that can vary continuously. So maybe overwrites leave you with something like 0.1s and 0.9s instead of 0s and 1s?

According to this Cambridge paper about whether you can sanitize flash memory to purge overwritten data, the answer varies wildly depending on who makes the device. Most of the attacks involved taking the chip out of the packaging. Some were easier timing or power-glitch attacks. All the attacks are getting more difficult to carry out with each new generation of flash chips.

If you're seriously worried about having old data recovered by someone you don't like, the preventive measures are way different from those you'd use on a hard disk. For example, you should write to cells before the erase cycle rather than depending on repeated erase cycles. The paper warns "From some samples, information can still be recovered after 100 erase cycles.".

It's a highly technical paper, unless you already know what Fowler-Nordheim tunneling is.

How does spyware install itself?

2006-12-16T03:02:00.000-08:00

Ben Edelman's list of spyware installation techniques

There's still no excuse for all the drive-by downloads possible with Internet Explorer, but notice how few of the vectors are exploits and how many are bundles and social engineering.

Slashdot summary, things to do about the unpatched Word vulnerabilities

2006-12-16T02:39:00.000-08:00

jayjay converts Word documents to PDF at the server.
slamb suggests using a version control system for collaborative editing, PDF for finished documents, and then banning Office documents from email attachments.

You thought maybe financial data transfers were encrypted?

2006-12-16T02:34:00.000-08:00

In case you missed this on Slashdot, a programmer who works on electronic funds transfer and card authorization systems says merchants send batches of credit card data in plaintext over POTS lines.

It's not just phone taps that could put that information at risk. There are other ways to get data out of a modem.

Detail on the latest Word vulnerability

2006-12-14T18:02:00.000-08:00

"Data used by Microsoft Word to construct a destination address for a memory copy routine is embedded within a Word document itself. If an attacker constructs a Word document with a specially crafted value used to build this destination address, then that attacker may be able to overwrite arbitrary memory."
My brain hurts. Either this statement is trivially vacucous (the whole idea of file formats being to fill things into memory locations) or it reflects a truly bizarre design choice.

Scramble your VOIP calls

2006-11-15T20:14:00.000-08:00

Phil Zimmermann's (yes, that Phil Zimmermann) ZFone encryption plugin is in public beta. It works with SIP-based systems but is not SRTP. Designed to be lightweight and opportunistic.

It also hooks into your network stack, so be sure you have a way to back out of the installation.

Mac and Linux clients encrypt video.

Need to explain to someone why ECB is bad?

2006-10-28T15:15:00.000-07:00

From an MSDN article about the software development lifecycle, a vivid illustration of how ECB doesn't protect you well.
Plaintext, in the form of an image file
The same image encrypted in ECB mode
CBC mode

Hazards of copyright lawsuits?

2006-10-27T19:01:00.000-07:00

There's no convenient way to check this story, but an anonymous poster on Slashdot reports that a **AA investigator installed malware on a computer under investigation.

The story is missing some details and coherence that I'd like to see before accepting it with confidence.

Top ten user errors -- it looked good at first

2006-10-27T12:59:00.000-07:00

Top ten most dangerous things end users do online

It's from a corporate point of view, which means that their idea of security is that employees shouldn't surf porn sites because the company might get sued for allowing a hostile working environment. If your worry is keeping the computers running and secure, the problem is that too many porn sites are bait to lure people into installing malicous software.

They say "Opening HTML or plain-text messages from unknown senders" is dangerous. No. Just HTML.

The rest of their advice is sorta OK, though if you follow all of it you'll miss out on some of the most valuable things you can do online. Most of the warnings have to do with avoiding exposure to attacks against your web browser. So just choose one for security, like Firefox or Opera. (I really like what I'm hearing about the design changes under the hood of IE7, by the way. It could turn out well).

In case you missed this: if all you have is DNS,

2006-10-26T21:52:00.000-07:00

then you can still send general IP traffic. The "iodine" package tunnels IPv4 through a DNS server.

Vista sorta fixes bypass of driver signature checking

2006-10-24T17:19:00.000-07:00

At this year's Blackhat, Joanna Rutkowska pointed out that even though the Vista kernel refuses to load unsigned drivers, the kernel code can get paged out and administrator-level programs can edit the page file. She showed the details of how to make it work.

Microsoft has turned off raw disk write from usermode, which she thinks is neither necessary nor sufficient (and I can't figure out why it was possible in the first place).

More comments at Rutkowska's blog.

Need hard data about fingerprint security?

2006-09-10T00:21:00.000-07:00

Of course the scanners get fooled by artificial fingers. The only wayto get repeatable results in testing is to use artificial fingers. Though it's still hilarious to read about over-hyped systems getting fooled by someone with access to Gummi Bears.

If you want information that's freer of ax grinding than what you see in the press or in marketing materials, check out http://fingerprint.nist.gov/minex04/index.html.

Be wary of online password checkers

2006-07-14T19:18:00.000-07:00

A password quality checker from Microsoft and another password quality checker from HANetworks are today's topic.

Both of them think "HeavyMetal1980" is a strong password. I picked that one because I remembered reading about it connection with a web defacement case, back when people still did that. A cracking program uncovered it. It's a weak password or at best a mediocre one.

Both of them think "cleft cam synod yr" is a weak password. That one is a Diceware passphrase. Since the words are chosen at random from a pool of 7776 possibilities, Diceware provides about 51.6 bits of entropy per 4-word passphrase. The passphrase is one of 3.66 quadrillion equally likely phrases. Brute-forcing that would take serious computer time even on today's machines. It's a lot stronger than a common phrase with a calendar year stuck on the end.

Why do the quality checkers make such silly mistakes?

Partly because they're mistaking the means for the end. They're looking for mixed character sets, for example. Those are good things, for sure, but a few minutes of quality time with a spreadsheet should convince you that a password's length matters more than what you do with it.
They're apparently not hooked up to cracking dictionaries. Nobody wants to spend server CPU on a free site to run John the Ripper, of course, but they could at least have a fast search through a static dictionary. That would save them the embarrassment of classifying the phrase "Pamela Anderson" as a medium-strength password.
The problem is impossible to solve. Entropy is a property of the generator, not of the output, and isn't measurable given a single output. You could substitute the metric of time-to-crack-with-program-X, but then you'd have to wait a long time for the results.

Second preimage attacks on HMAC

2006-07-10T17:30:00.000-07:00

HMAC said to be disntiguishable from pseudorandom, forgeries possible.

This is confusing: Google searching inside binaries?

2006-07-09T00:19:00.000-07:00

We knew that Google crawls inside PDFs, Word documents, and Powerpoint slides. Websense says that Google is alson looking inside executables and that it's possible to search for malware signatures.

Fascinating, but they don't say what the Google search syntax is, and there's nothing obvious in Google's documentation.
UPDATE 7/11:
A litte more information at http://googlesystem.blogspot.com/2006/06/some-google-results-are-exe-files.html