Slashdot Mirror
Getting Hacked Through Your Terminal
hdm writes "My company recently published a paper on security issues with common terminal emulator applications. The interesting thing about these vulnerabiltiies is that many of them only require the victim to be running tail on their log files (apache, syslog, etc) for the attack to be successful. The paper (TXT) can be found here."
Most exploits that existed over 5 yrs ago are still valid.
For instance, you can recover a root password in just 10-15 minutes on ANY machine.
31 people regularly point & click my G-spot
just pipe your stuff through strings. it should help a bit
Given the profliferation of exploits related to race conditions, predictible file creation, etc,
;-)
we should henceforth re-tool our code to only make use of stateless protocols!
The difference between stupidity and genius is that genius has its limits.
I always pre-filter my logs thru a Perl script. Besides removing verbose messages that are not useful, my filters replace such non-printable characters with a printable character. Not only does it make it easier to spot strange octets on the screen, it does not depend on the terminal emulator remaining secure.
chongo (was here)
OK, so the exploit descriptions seem real and interesting enough, but what was the point of such an over-the-top
"Fictitious Case Study"?
Reading that drivel about Andre and his l33t hax0r buddies totally removed the credibility the rest of the article had achieved!
People would know what I mean't if I wrote in l33t, but it would be a stupid thing to do.
So if it was in HTML the link should have been "the paper (HyperText Markup Language)..."?
Come on, is it that big of a deal really?
It's a text file on what appears to be a decent server, not some Joe Q. User's geocities account, discussing a topic of relatively low interest to most people.
In other words, it's not going to get slashdotted, so stop karma whoring.
A Minesweeper clone that doesn't suck
So they discovered ANSI bombs over again.
Simple! Just tell Linux not to load ANSI.SYS, problem solved!
Another day, another vulnerability. These exploits are getting more bizzare and more useless every day. The risk factor here is ridiculously low.
I don't want to here about anymore useless exploits or no risk vulnerabilities. If you really want my window title, I'll telll you what it is; Getting Hacked Through Your Terminal - Konqueror
Now, when someone gets an exploit to replace the Slashdot ads with Goatse, then I'll be impressed.
However, certan software would allow an attacker to insert these control codes anywhere, and not just interpret them from the menus.
Imagine the hilarity that insues once the attacker figures out the embed sequence for the drop to DOS feature.
The more you know, the less you understand.
Someone has to be using a terminal for someone to be able to do this to them. Is it just me, or is the solution really obvious? Just chmod every thing with a command line to 000! That should keep those naughty, naughty crackers out!
Note to M1-ers: a curt but otherwise insightful message is not "Flamebait" or "Troll".
Apparently the OSX terminal might be susceptible to this. It is possible to alias different escape sequences to commands like lm and ll to make the terminal full screen, send it to the background, make it tall, etc.
/var/log/messages, but I won't be the one to find out as my PowerBook is in the shop...
It would be interesting to see whether actual commands can be executed if tailing
Back in the day, "Ansi Bombs" were considered an art form. With the art scene so active, you could usually embed some evil escape string in a good looking graphic and know that you were going to get people.
The problem was DOS' overly-powerful ANSI.SYS interpreter. It let you remap any key to an arbitrary set of keys, making keyboard macros pretty easy. However, it also let evildoers remap "Space" to, for example, "del *.*, enter, y, enter." Luckily, there were third party ANSI interpreters that didn't suffer this vulnerability.
One time, when I was about to reformat my HD, I even wrote an ANSI bomb to do it. Crazy stuff. There's an interesting (and of course, old) paper about it here.
It all goes downhill from first post
I dare to say that putty IS vulnerable. This is what happens when I run a similiar command. Sure, it still expects an enter and anyone who takes his time to read stuff on his/her screen before blatantly hitting enter will notice the command. This was done with the latest build which I just downloaded from the site.
Hate me!
If you had read the article, rather than just glancing through it... those were the terminal emulators he used.
Eterm was affected, Putty, Xterm, and Rxvt.
Black and grey are both shades of white.
So they discovered ANSI bombs over again.
Looks like it.
I recall the "ANSI Standard Trojan Horse" (though it was misnamed). It became quite common after ANSI standardized a cross between the VT100 and Ann Arbor Terminals escape sequences.
Simplest form was to send a "talk" mesage to a root user containing an escape sequence that programmed a hotkey with your command-to-be-run-as-root and then "hit the key" by remote control. (You could even bury it invisibly in an otherwise innocuous-looking message.)
ANCIENT stuff. What-early '70s?
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
Most terminals released in the last few months aren't afflicted by this, particularly the most common. You have to be running a terminal to get hacked like this. You need some pretty contrived circumstances to get hacked like this; your average opportunistic script kiddie won't be able to use this trick too easily on strangers. Besides, couldn't you just set up a cron job to pipe all the logs through strings?
Lonely man seeks satisfying cock. My password is 'ha
In my neck of the woods TXT is practically synonymous with text messaging. No, actually it's, synonymous with the delivery of TXT msg svrl hrs aftr u snt thm...
# init 5
Connection closed.
Oh...
The author went on to have an interesting converstaion with Micahel Jennings, author of Eterm on Bugtraq here.
ex$$
The following terminal emulators were found, according to the article, to NOT be susceptible to screen dump or window title attacks:
Some asked about my Perl filter for tailing log files.
Sans typos, here is an example that removes certain types of messages and fields, checks the file every 60 seconds, picks up trailing on the new file when the log file gets rotated (moved away), trims to 224 characters and replaces unusual chars with ~'s (assuming you use ASCII).
As they say in perl, there is more than one way to do it. The above code fragment is just to give you the general idea.
chongo (was here)
Actually, Windows *does* have a logging feature, but, in typical Microsoft overdesigned-anal-retentive fashion, it's a pain in the ass to use. Look at "Event Viewer" in the Administrative Tools sometime. I forget the API, but you (the programmer) need to fill out some complex structures and do something-or-other unintuitive to dump a message to the log. To actually see it, you need to go through a scrolling list of messages marked "Information" and doubleclick one to see the actual text.
I don't think anyone actually uses this, other than driver writers and those with the plugs in the backs of their heads.
What if life is just a side effect of some other process and God has no idea we exist?
The paper mentions injecting escape sequences into log files which are being tail -f'ed... and that there's nothing new about terminal exploits.
/var/log/apache/common.log' would limit less to 1024K (1M) of buffer, which means old data will eventually be discarded, but keeps less from malloc'ing all your core. As always, Read The Fine Manual for details :)
When I first heard about this (a couple of years back) I started using less +F for tailing logs. less will convert the escape character into the token ESC (in bold or inverse video), avoiding any escape-sequence exploits.. and also adds the benefits of being able to scroll back and search, which would make it worth using even if there were no such thing as a terminal exploit.
If you're going to leave it running for a long time, you might want to also look into the -b and -B options, to limit the amount of buffer space it will allocate: something like `less -b1024 -B +F
I just checked: the `more' command on my Linux and Solaris boxen seems to pass escape sequences through, so you really do want `less' (or alias less=more, if you're used to typing `some_command|more'), not to mention `more' has no equivalent to `less +F' or `tail -f'.
Hope this helps someone...
One point that stood out at me is how many apps are vulnerable because they started with an insecure codebase. This is how exploits become common - one buggy piece of code spreads to a dozen others and never gets fixed in any of them, if only because authors say "let's add our new features instead of validating these 25,000 lines of code we started with."
What if life is just a side effect of some other process and God has no idea we exist?
By trimming down to 224 characters, I avoid line wrap. System crackers cannot cause a very long line to wrap at just the right place to fake another entry.
Sure, I'll perhaps miss important text beyond the 225th char, but I'm willing to take that risk. Besides, the script converts tabs to a single space and removes some common fields and phrases which further cut down down on long lines. And I can always to back to the logs to look at really long lines if I want to.
And yes, maybe some terminal emulator has some long line bug somewhere. My perl script helps avoid testing it. :-)
chongo (was here)
Some years ago I managed to make an xterm dump core simply by typing the contents of a binary file, which did not contain any malicious data, it just be coincidence contained a byte sequence that would cause xterm to dump core. Investigating this I found, that some sequence of 24 bytes was responsible. But if I made the xterm window larger, it would not crash from this sequence, however it turned out the original file contained another longer sequence that could also make a larger xterm dump core. I never actually understood exactly what was going on, but I found that later xterm versions does not crash from the sequences I saved from back then.
Do you care about the security of your wireless mouse?
I have checked your journal, and there are no links to porn sites. Please rectify this immediately.
Thank you,
Anonymous Coward
will take care of the tail of a log file problem...
In the article, especially in the 'Fictitious Case Study', the author makes quite a lot of snide (although funny) remarks about Enlightenment.
By the way, I am not an Enlightenment user, so don't think that is why posted this.
For instance...
...or...
Thank you.
GrimReality
2003-03-02 01:27:02 UTC (2003-03-01 20:27:02 EST)
Similar situations existed for the registry, DCOM, performance counters, security model, etc., etc.
How many of us happen to put pictures of our terminal windows up on our websites?
A long long time ago, on a uucp-net far, far, away, we didn't use terminal emulators, we used real hardware terminals. Most of them didn't run ANSI (I don't remember if the ANSI terminal standards were defined yet, but they looked very much like a DEC VT-100 terminal), but that meant there were lots of types of terminals, all trying to achieve market differentiation by having cool features or small size or large screens or cool plastic cases. Lots of different cool features means lots of vulnerabilities. The hackers in question had REDISCOVERED AN OLD TECHNIQUE - they found ways to hand escape sequences to VT100 terminals that would get the terminal to send arbitrary text back to the computer, which in those good old command-line days meant they could do anything they wanted. All you needed to do was email somebody a message with carefully crafted character strings in it, or use the talk program to write them to their terminal (back when everybody left their ttys writeable so talk would work.) Some of the popular techniques were to abuse programmable function keys (send a sequence to put some string in F1, then another sequence to auto-trigger the F1 key), or automatic whoami-responders (VT100 had one of these), or put the terminal into loopback testing mode (letting you type anything you want.)
The Hewlett-Packard HP2621 and its relatives had two easy things to do, which didn't let you send arbitrary characters but still hosed the user - one did a reboot on the terminal (and thereby dropped the connection), while the other put it into test pattern mode (sending a long string of UUUUUU to the computer.) There may have also been some block-mode things that let you send arbitrary characters to the computer, but these two were the easiest and best-documented. A couple years later, when I was a newbie learning Unix programming and security at Bell Labs, and Robert Morris Senior (father of RTM the Worm Author) was a department head for computer security (before he went to NSA), we'd had a discussion about some techniques I was trying (which he cracked through personally :-), and a few days later, when I'd patched those holes, I was playing rogue in the evening, and one of his people talked the test pattern sequence at me. For a few seconds, I though Umber Hulks were suddenly going to eat my character, but then realized I'd been hacked, and it was one of RTM's people following up on what I'd patched and what I hadn't...
Terminology note on "hacker" as opposed to "cracker" or "vandal" - The folks at Berkeley really were hacking - tinkering with things to find their limits - and while they could potentially use their knowledge for evil, they instead told people about it, which reminded lots of us to check for potential security holes in our systems.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
The Great Hacker (later read: me) would dig up docs on VT10x escape sequence interpretation. Then I would look to see what that trimmed down file translates to in terms of screen manipulation code.
I would think to myself, does that series of actions even make sense? It probably wouldn't (seeing as it tends to crash xterm).
The next thing would be to trim the sequence down to just escape codes (removing any printable characters in the sequence), and see if still dumps core. If not, its probably due to a simple segfault (whether writing character or blitting a bitmapped font in an invalid buffer).
Then try removing an escape code or chaning a parameter. Does it still dump core? Are there ranges of values for each escape code that cause the behavior?
Finally, you'd grab the source code and see if the are any obvious flaws. Does the emulator load a parameter and then cast it into a signed value? Then does it not check to see if it was negative? etc.
All those steps (which I've never done nor considered before your post) would be my plan to handle and investigate the situation.
I think what it requires to be a Good Hacker, therefore, is a sick motivation to get to the bottom of things, no matter how painful or time consuming the excercise. The steps I outlined above could easily take 15 hrs. to adequately address. Are the potential results worth that time? To a great hacker, who relishes such experiences, yes they are, even if she cannot pinpoint the problem or fix it.
There was nothing that you did wrong. I think maybe you didn't really want to be one. Great hackers do not just have fits of ingenuity and shit. They just hack at things until it becomes clear. It's time consuming, tiring, and often thankless. But over time, you get to have a certain knack for relating new problems to old ones, and it becomes easier.
Nothing says you can't start now either. Go for it!
Fuck Beta. Fuck Dice
What if the computer is like my own, where I have a BIOS password *and* a padlock locking the case?
And here I thought they were referring to RFC 1464: Using the Domain Name System To Store Arbitrary String Attributes....
Larry Wall's famous quote says Great Hacker virtues include Laziness, Impatience, and Hubris.
I would agree, and say that a Great Hacker would locate those killer ctrl codes, get a boatload of diagnostic information together, and hand all that over to the maintainer of the terminal program and then not sweat trying to figure it out themselves.
Let someone that knows the code, and might be better able to divine what's up have a wonk at it. You could even check the diffs if it ever gets patched, and you were curious and had time to kill.
Then, the Great Hacker would go and patch their own project to fix up the newest incoming bug report.
Then, the Great Hacker would rest and eat of the Divine Pizza.
And all would profit.
Keep your packets off my GNU/Girlfriend!
You do have "mesg n" set, do you?
If not, you are attachable by "write " as well.
And yes, this is very, very old. Many an administrators open terminals have been taken remotely using this when I was at university.
Kristian
tail -f /var/spool/foo/logfile | cat -uv
Something any smart sysadmin would do, anyway, to display just exactly what was showing up in the log.
Some programs or setups spit stuff out to the console - warnings, error messages etc.
I remember doing something similar when testing my company's TIS FWTK setup years ago. When you used an address with unbalanced brackets, sendmail when called by smapd will grumble to the console. As a test I sent an email with a ESC sequence setting the foreground colour to black and it worked. Wasn't too surprised. Wasn't able to find an exploit for the console term at that time, and no one abused it (security through obscurity).
Later I switched to using qmail (IIRC qmail 1.0.3 dates to 1998, the TIS FWTK has been around for quite some time). Just to test things out I also recently wrote an smtp proxy to enforce state and filter out stuff(qmail may be robust, but it tends to pass the junk unfiltered to other stuff which may be less robust).
Some O/Ses send a many kinds of messages to the console too.
But the later option was too risky for my taste, because the "login" process was owned by you. So instead, I wrote a doomsday shell script. It gave you a # prompt to make you believe you are running as root. It then emulated various UNIX commands. For example "jobs" showed [1] rm -rf / & and "kill" returned "Permission denied". It logged all the commands and responses to an obscure file in my home directory. Once I got our semi-knowlegable system administrator's assistant to "interact" with this script and it was quite fun reading him using kill 10 times in a row with the same arguments. She really thought the filesystem was going south.
Terrible abuse that can be inflicted on X terminals or public lab PCs with terminal emulators is left to the imagination of the reader.
You need to compile a messages file using the mc.exe message compiler, then rc -r the output to a .rc file.
.rc file into the executable or into a .dll (for language specific error messages etc).
You need to regster the event source and either link the
Damn, I remember being a Perl & Unix hacker, hot the fsck did I end up coding for Winblows XP???
Rich
This story gives a whole new meaning to that phrase.
I hereby place the above post in the public domain.
Oh god, that's right. It's all coming back...
I remember looking at that whole business once and going back to logfiles. Although, the message/resource file business *is* a lot more maintainable and internationalizable, and the process isn't all *that* bad. Let's just say it's overkill for 90% of things.
Windows - the Ada mindset of software development!
Damn, I remember being a Perl & Unix hacker, hot the fsck did I end up coding for Winblows XP???
That nice big paycheck at the end of the week helps mitigate the pain and guilt, I've found. Of course, "big" and "mitigate" are relative words.
What if life is just a side effect of some other process and God has no idea we exist?
The pay helps, a lot, but hey, one day I might just take up mowing lawns for a living.
But I did just install XP of a 2MB per inute network link - I feel dirty... *grin*
I think that's nonsense.
I'm certainly not professing to be a great hacker, and I don't think people in that category would use that term to desribe themselves.
However...
Not bothering to spend the time working out why a given string crashes an xterm does not mean you are not a great hacker, let alone that you will never be one.
All it means is there was something else you would rather do instead. Which is not a bad thing.
Obviously, you can't fix every software bug in the world on your own - and clearly the best way to deal with them is not to try and fix them all personally, so another approach is needed.
To put it another way:
If everybody spent time trying to fix obscure bugs in other people's code while still trying to write their own software, all development would grind to a halt - developers would spend loads of time pouring over and trying to understand documentation for libraries and API's and going line by line through other people's, very possibly krufty, code - instead of spending time working on solving their own bugs.
If I found such a bug, I'd be likely to contact the author informing him (and provding some details as to why), and go back to fixing *my* bugs.
Although it's no bad thing to debug and fix bugs in other poeple's code if you have some free time and no projects of your own, it's simply far more logical and efficent to spend time fixing your own bugs.
PS: FWIW, IMHO, 'it' is simply equal lashings of time and effort and a dollop of hubris to taste, served on a bed of sufficent basic intelligence. The 'time' and 'effort' ingredients being the hardest to come across.
Lines such as:
serve as part of the ''removes certain types of messages'' processing. And lines such as:
serve as part of the ''removes certain types of ... fields'' processing.
And lines such as:
serve as part of the ''trims to 224 characters'' processing. And lines such as:
ensure that each message read is printed with a trailing newline.
To protect terminal emulators, the following lines are used:
For systems using ASCII (and a number of other widely used character sets), it replaces potentially dangerous characters with ~ (tilde) characters. The replacement is 1-for-1, so the layout of the original message is mostly preserved.
I say ''mostly preserved'' because a tabs are replaced with space. I prefer to view whitespace as just token separators and to reduce them where reasonable. A more complete whitespace crusher would use:
On the other hand, you might want to keep all whitespace as it is when tailing log messages in your terminal emulator window, so:
would do the trick.
And of course one could write the regexp to not depend on a character set order. I'll leave that exercise up to the reader. :-)
BTW: I prefer to replace strange characters instead of tossing them. Tossing strange characters instead of replacing them could would allow a system cracker to potentially mimic a valid messages knowing that invalid chars would simply be tossed. And I usually want to know when strange chars appear in the logs.
I hope this helps.
chongo (was here)
I just tried this in GoGlobal 1.63:
bash$ echo -e "\e]2;This is the new window title\a"
2;This is the new window title
Dunno, maybe it's a different code number.
Somebody let me know...
YMMV
- OrbNobz
Finding more and more people obsequious, obstreperous, and ubiquitous.