U.S. Navy Works To Improve Linux Security

MrPhiles writes "Just saw an article at Washington Technology talking about how the Navy is developing a Secure Auditing tool for Linux. I think it's cool that government agencies are taking steps to obtain credentials necessary for open source use in high-security environments."

Comments

Great but

[jsse]

I wish they'd spend more money on auditing Windows too.

Of course, crash on "division by zero" is a feature, not a bug. :)

Re:Great but

[rritterson]

It's not quite the same, obviously. The Navy software engineers can easily tweak and fix any holes they find. With Windows you are limited to the framework MS provides and the hope that they will fix any problems discovered.

The NSA released documents on how to secure WinXP and Win2K server not too long ago- it was even posted on /.

Re:Great but

[Repugnant_Shit]

I still need to plow through that PDF. It looked like it a lot of good info.

Re:Great but

[4of12]

The NSA released documents on how to secure WinXP and Win2K server not too long ago

I'm glad they did that. It was nice public service, IMHO.

However, for practical use, the 105 page guide is a bit prolix for me. I'm installing, patching and trying to harden a home Win2K system (got removable drives and SuSE 8.1 on the other) and found other, shorter guides (ArsTechnica, I think) for Win2K security to be quicker and easier to use.

what is the kernel lacking?

[zogger]

I see in the article that the linux kernel "lacks" such and such for security auditing? Would one of ya'all gurus please explain this? I thought there were a plethora of auditing tools and schemes already. Thanks in advance!

Re:what is the kernel lacking?

[Beryllium Sphere(tm)]

>Would one of ya'all gurus please explain this?

Attend, my son :-)

The key word seems to be "forensic". They want to replace syslog with something sufficiently tamper-resistant to persuade a judge that it's good enough for legal evidence. There are already some clever hacks for this, such as hiding the real syslog process and leaving a fake one around for an intruder to disable or corrupt.

Re:what is the kernel lacking?

[ctr2sprt]

There are a lot of things that can be meant by "auditing." At its most sensitive, you can audit all accesses to certain system resources (files, syscalls, device node accesses, that sort of thing), and at a more standard level you can audit failed accesses to system resources. Some of this is implemented in some way - "bad root login on tty1" - but usually when people talk about system security auditing, they are talking about a unified framework for controlling auditing of the entire system (the parts relevant to security, at least). The "unified" part is especially important because frequently you want to analyze the audit trail using automated tools, kind of like Snort and ACID, rather than viewing each entry individually. If you have 50 different programs auditing things in 50 different ways, that makes system-wide analysis much more difficult.

This isn't the sort of thing normal people will want or need on their Linux systems, but in some environments (military, government) it's really important. And you could, of course, use it to create a honeypot, if you're into that sort of thing.

Re:what is the kernel lacking?

[zogger]

--ah, I was not aware that such a tool didn't already exist. Being a still neophyte at this I am still learning various tools. Making "one" tool that *does it all* seems logical, except for the single point of failure phenomenon then.

Of course, you are correct, most "normal" users don't seem to need this. In fact, as a "normal" user, I must say I certainly...uh.. enjoy... all the "volunteer" efforts that kind hearted "outside auditors" seem to be always giving me... uhh ya... enjoy..... I guess.....

%^)

Re:what is the kernel lacking?

[bill_mcgonigle]

They want to replace syslog with something sufficiently tamper-resistant to persuade a judge that it's good enough for legal evidence.

Just echo the syslog output to a 9-pin dot matrix printer...

Re:OpenBSD

[bonsai_kitty]

I couldn't agree more. *BSD would be better at this task IMHO. Yes the Development people can be fickle but isn't the government as well? ;)

Re:OpenBSD

[gearheadsmp]

DARPA dropping the OpenBSD grant just goes to prove that the U.S. Department of Defense cares more about their self-image than security & stability. To me, this makes them appear to have swapped priorities with the White House. Instead of focusing on Defense, the DoD is focusing on PR. Maybe the Canadian Defence Ministry will pick up where DARPA left off.

Re: with no desire to be clever

[Black Parrot]

> navy penguin

That's the guys who weren't quite tough enough to make the Seals, right?

ok...

[zogger]

..ok, that makes sense. so in order to do that, following normal procedure (made infamous in the OJ case) you need a provable uncorrupted "chain of evidence" from start to finish.

Turbocharged DRM would of necessity be part of that along with the allegedly "incorruptable" logs. It matters now what you are looking at with regards to this theoretical 'crime" if the evidentiary analysis would not be able to prove a "perp". Proving the crime occurred seems to be the premise of the hardened logs, but proving who did it is still ellusive WITHOUT mandated suber turbo DRM styled efforts.

Or so it looks like to me.

How is this different than the NSA's SE Linux?

[cmehta1]

Are the Navy and NSA working on the same kinda things? Or do we have more govt waste with duplication efforts?

And just to get more tweaky...is it also similar to the aborted Dept. of Defense changes that Theo de Raadt was gonna do on BSD?

NSA page: http://www.nsa.gov/selinux/

Great but

[__aaimgy6840]

Why on earth would the Navy spend good money auditing Linux, when OpenBSD is already the most secure OS? It's been audited for the last 6? years.