Feedback: Please send feedback to "email@example.com".
Each word is marked with a number that describe how "technical" the
word is. The following scale is used:
 used by people inside and outside the community
 common terms used within the community
 rarer terms used in technical discussions
 rare terms used by experts; these terms will require extensive study in order to really grok them.
 rare terms used by virtually nobody (i.e. if you are using this guide to learn the lingo, skip these terms).
Copyright 1998-2000 by Robert Graham (firstname.lastname@example.org. All rights reserved. This document may be reproduced only for non-commercial purposes. All reproductions must contain this exact copyright notice. Reproductions must not contain alterations except by permision..
Key point: The debate over strong encryption is never ending. Within the United States, law enforcement is constantly lobbying to restrict the use of strong encryption. Many resist, pointing out how often law enforcement already abuses wiretap powers (such as against Martin Luther King). At the same time, companies making products constantly lobby for the easing of export restrictions, so that they can sell strong encryption products abroad. Another funny thing is that the U.S. government's intransigence on this issue has actually led to stronger encryption abroad. U.S. export restrictions (and desire to spy on foreigners) was one of the reasons France relaxed its own law-enforcement bans on encryption use by citizens.
Key point: Specialized hardware can decrypt 40-bit keys in real time. The average new desktop has enough horsepower to decrypt 40-bit messages. Thus, many people now consider 40-bit encryption to be simply obfuscated plaintext.
Key point: 40-bit often refers to the RC4 system within browsers.
Key point: In January of 1999, the EFF built a custom machine (the "Deep Crack") for $250,000 that could decrypt 56-bit DES encrypted messages in hours.
Key point: 56-bit cryptography almost always refers to DES.
Key point: Security conscious users of such systems need to make sure they use a more random mix of characters because they cannot create long passwords.
Key point: Password cracking such systems is a little easier.
Key point: Web-servers often allow access to user's directories this way. An example would be http://www.robertgraham.com/~rob.
Key point: A big hole on the Internet is that people unexpectedly open up information. For example, the file .bash_history is a hidden file in a person's directory that contains the complete text of all commands they've entered into the shell (assuming their shell is "bash", which is the most popular one on Linux).
Key point: When rooting a machine,
hackers will often redirect logging to /dev/null
For example, the command
Culture: In the vernacular, means much the same thing as "black hole". Typical usage, "if you don't like what I have to say, please direct your comments to /dev/null".
Key point: If a hacker can read files from this directory, then they can likely use the information to attack the machine.
Key point: The passwords are encrypted, so even though everyone can read the file, it doesn't automatically guarantee access to the system. However, programs like crack are very effective at decrypting the passwords. On any system with many accounts, there is a good chance the hacker will be able to crack some of the accounts if they get hold of this file.
Key point: Modern UNIX systems allow for "shadowed" password files, stored in locations like /etc/shadow that only root root has access to. The normal password file still exists, minus the password information. This provides backwards compatibility for programs that still must access the password file for account information, but which have no interest in the passwords themselves.
Key point: The chief goal of most hacks against UNIX systems is to retrieve the password file. Many attacks do not compromise the machine directly, but are able to read files from the machine, such as this file. Typical examples include:
Key point: An Access Control List (ACL) is used to list those accounts that have access to the resource that the list applies to. When talking about firewalls, the ACL implies the list of IP addresses that have access to which ports and systems through the firewall. When talking about WinNT, the ACL implies the list of users that can access a specific file or directory on NTFS.
Contrast: Discretionary Access Control is the ability to have fine grained control over who has access to what resources.
Analogy: An cookbook recipe is an algorithm.
Key point: Different algorithms have different levels of complexity. For example, consider the ancient parable (Babylonian?) about a king and a wise subject who did a favor for him. The subject asked for one piece of grain to be placed on the first square of a chess board, two grains on the second, four grains on the third, and so on, doubling the amount of grain for each successive square.
This problem demonstrates an algorithm of exponential complexity. For the first 10 squares of the chess board, the series is: 1 2 4 8 16 32 64 128 256 512. Thus, for the first 10 squares, roughly a thousand grains must be paid out. However, the series continues (using K=1024): 1k 2k 4k 8k 32k 64k 128k 256k 512k. Thus, for the first 20 squares, roughly a million grains must be paid out. After 30 squares, roughly a billion grains must be paid out. For 40 squares, roughly a trillion grains must be paid out.
This is directly related to such things as key size. A 41-bit key is twice as hard to crack as a 40-bit key. A 50-bit key is a thousand times harder. A 60-bit key is a million times harder. This is why the 128-bit vs. 40-bit encryption debate is so important: 128-bit keys are a trillion trillion times harder to crack (via brute force) than 40-bit keys.
Key point: Most algorithms are public, meaning that somebody trying to decrypt your message knows all the details of the algorithm. Consequently, the message is protected solely by the key. Many people try to add additional protection by making the details of the algorithm secret as well. Experience so far has led to the belief that this actually leads to weaker security for two reasons. First, such secrets always get discovered eventually, so if security depends upon this secret, it will eventually be broken. Secondly, human intelligence is such that someone cannot create a secure algorithm on his/her own. Therefore, only by working with a community of experts over many years can humans create a secure algorithm. To date, only two such communities exist: the entire world of cryptography experts publishing the details of their work and trying to break other people's work, and the tightly knit community of cryptography experts working in secret for the NSA.
Key point: Anonymous e-mail services like Hotmail put the IP address of the person sending the e-mail in the headers (which are normally hidden from view by e-mail clients). Many would-be hackers get caught this way.
Key point: A recurring bug in ASP has allowed hackers to read the script rather than the output of the script. These techniques rely upon changing the name of the script such that the server not longer recognizes it as a script, but as a file instead. Some techniques that have worked in the past have been:
Key point: By sniffing ARP packets off the wire, you can discover a lot of stuff going on. This is especially true of cable-modem and DSL segments. Since ARP packets are broadcasts, you aren't technically breaking your user's agreement by sniffing.
Key point: You can spoof ARP requests and/or responses in order to redirect traffic through your machine.
Example: Some classifications of attacks are:
Example: When you log in with your username and give the password, you are authenticating yourself to the system. You are proving that you are you because, in theory, only you know your password.
Key point: Abstractly, anything that combats forgery is called authentication. For example, IPsec includes an Authentication Header (AH) that proves that a packet hasn't been modified in transit.
Contrast: Remote access trojans (RATs) are NOT examples of back channels, but are instead forward channels. A RAT allows the hacker to contact the system from anywhere in the world, and allows the hacker to hide where he/she is coming from. A back channel, on the other hand, will contact the hacker, who must have a fixed IP address. This clearly fingers who the hacker is.
Key point: Typical back channel protocols are X Windows (xterm) and shells like Telnet. These programs are often built into the victim's system, so many attacks that can't otherwise compromise the system can still trigger a back channel that allows a remote shell.
See also: covert channel
Key point: Back doors are frequently programmed into systems either benignly or maliciously. Most computers shipped today allow BIOS passwords to be set that will prevent the booting of the computer without the administrator first typing the password. However, since many people lose their password, such BIOSes often have a back door passwords that allows the real password to be set. Similarly, a lot of remotely manageable network equipement (routers, switches, dialup banks, etc.) have backdoors for remote Telnet or SNMP. The frequency of such back doors is due to the fact that people are stupid, set passwords, forget them, then whine to customer support.
Example: Most backdoors consists of special username/passwords that can be entered in the normal locations where prompted.
Key point: A backdoor can be added to any system. For example, when generating random session keys, a programmer may actually subvert the random number generator. Such subversion would then allow decrypting of the message by those who knew the specifics. This has already been done accidentally; some paranoids believe that some encryption products do this intentionally in order to get export approval of 128-bit products.
Key point: Many banners reveal the exact version of the product. Over time, exploits are found for specific versions of products. Therefore, the intruder can simply lookup the version numbers in a list to find which exploit will work on the system. In the examples below, the version numbers that reveal the service has known exploitable weaknesses are highlighted.
Example: The example below is a RedHat Linux box with most the default service enabled. The examples below show only the text-based services that show banners upon connection (in some cases, a little bit of input was provided in order to trigger the banners). Note that this is an older version of Linux; exploits exist for most these services that would allow a hacker to break into this box (most are buffer-overflow exploits).
|FTP||21||220 rh5.robertgraham.com FTP server (Version wu-2.4.2-academ[BETA-15](1) Sat Nov 1 03:08:32 EST 1997) ready.|
Red Hat Linux release 5.0 (Hurricane)|
Kernel 2.0.31 on an i486
|SMTP||25||220 rh5.robertgraham.com ESMTP Sendmail 8.8.7/8.8.7; Mon, 29 Nov 1999 23:28:31 -0800|
Login Name Tty Idle Login Time Office Office Phone rob Robert David Graham p0 Nov 29 22:51 (gandalf) root root p1 Nov 29 23:34 (10.17.128.201:0.0)
HTTP/1.0 200 OK|
Date: Tue, 30 Nov 1999 07:34:59 GMT
Last-Modified: Thu, 06 Nov 1997 18:20:06 GMT
|POP3||110||+OK POP3 rh5.robertgraham.com v4.39 server ready|
|identd||113||0 , 0 : ERROR : UNKNOWN-ERROR|
|IMAP4||143||* OK rh5.robertgraham.com IMAP4rev1 v10.190 server ready|
|lp||515||lpd: lp: Malformed from address|
Key point: BGP can be subverted in numerous ways. BGP is generally unauthenticated, and rogue ISPs can play havoc.
Key pont: The BIOS stores configuration settings in NVRAM (Non-Volatile RAM). Remember that the contents of your normal RAM/memory are lost when you power-off your computer. The contents of NVRAM, in contrast, are retained when power goes off. Most NVRAM consists of CMOS (low-power) chips with a small battery that constantly feeds power to the chips (such batteries last about 5-years). A common trick of hackers and viruses is to corrupt the CMOS settings causing the computer to fail to boot. Removing the battery connection (usually a jumper on the motherboard) will cause the CMOS settings to be lost and be reset back to default (good) state.
Key pont: All of today's BIOSes are stored in programmable ROMs, which allows them to be reprogrammed (usually with bug fixes from the manufacturer). This allows the hacker to reprogram them as well. While in theory the hacker could reprogram his/her own code into the BIOS, in practice this has not been done yet. Instead, hackers can sometimes use this programming feature to corrupt the BIOS code (in much the same way they corrupt the BIOS settings mentioned above). This will usually prevent the system from booting even to a point where a fresh BIOS can be re-programmed into the system. This requires that the system be brought back to the vendor in order to have the BIOS reprogrammed. Note that you can often set a jumper on the motherboard that denies the ability to reprogram the BIOS.
Key pont: In many contexts, each additional bit means "twice as much". 8 extra bits means 256 times as much. 16 extra bits means 65536 times as much. Therefore, it takes 65536 times longer to brute force crack a 56-bit key than a 40-bit key.
Key point: DHCP is simply an extension on top of bootp. This is important because without an IP address, clients cannot reach bootp servers that reside across routers. Virtually all routers have an extension for bootp forwarding that fixes this issue. Since DHCP had the same requires, the designers just stuck it inside bootp packets rather than requiring yet another change to the routing infrastructure.
Example: A cancel-bot is a program that attempts to cancel lots of messages within USENET newsgroups. These are sometimes used by the USENET Death Penalty or rogue cancelers. *
Key point: Netscape and Microsoft have not yet produced a browser that is hardened against predation from hostile websites.
Analogy: If you somehow steal somebody's ATM card, you could try to use it in a bank machine. PIN numbers are only 4 digits, meaning 10,000 possible combinations. If you were patient, you could stand at the cash machine trying all possible 10,000 combinations. (Of course, ATM machines will always eat the cards after a few unsuccessful tries in order to stop this).
Key point: The term brute force often means "the most difficult way". In the above example of the PIN number, you can always find the PIN number after guessing 10,000 combinations. But sometimes there are easier ways. For example, a bank may choose to assign PIN numbers based upon a combination of the issuing date and the user's name. Therefore, the problem is reduced to guessing when a card was issued, which may consist of only a few hundred guesses.
Therefore, any technique that is more difficult than brute force is pointless. Likewise, brute force is very difficult, so hackers continually search for techniques that are less difficult.
Key point: The possibility of doing brute-force key-space searches is often compared to the age of the universe, number of atoms in the planet earth, and the yearly output of the sun. For example, Bruce Schneier has calculated that according to what we know of quantum mechanics today, that the entire energy output of the sun is insufficient to break a 197-bit key.
Analogy: Consider two popular bathroom sink designs. One design is a simple sink with a single drain. The other design includes a backup drain near the top of the sink. The first design is easy and often looks better, but suffers from the problem that if the drain is plugged and the water is left running, the sink will overflow all over the bathroom. The second design prevents the sink from overflowing, as the water level can never get past the top drain.
Example: In much the same way, programmers often forget to validate input. They (rightly) believe that a legal username is less than 32 characters long, and (wrongly) reserve "more than enough" memory for it, typically 200 characters. However, they frequently forget to check if some malicious hacker enters an illegal username 1000 characters long.
Key point: This is a classic programming bug that afflicts almost all systems. The average system on the Internet is vulnerable to a well known buffer overflow attack. Many Windows NT servers have IIS services vulnerable to a buffer overflow in ".htr" handler, many Solaris servers have vulnerable RPC services like cmsd, ToolTalk, and statd; many Linux boxes have vulnerable IMAP4, POP3, or FTP services.
Key point: Programs written in C are most vulnerable, C++ is somewhat less vulnerable. Programs written in scripting level languages like VisualBasic and Java are generally not vulnerable. The reason is that C requires the programmer to check buffer lengths, but scripting languages generally make these checks whether the programmer wants them or not.
Key point: Buffer overflows are usually a Denial-of-Service in that they will crash/hang a service/system. The most interesting ones, however, can cause the system to execute code provided by the hacker as part of the exploit.
Key point: The language is quirky, difficult for beginners to learn, and really just an accident of history. Despite this, one must grok the language in order to become a true hacker.
Key point: The large number of buffer overflow exploits is directly related to poor way that C protects programmers from doing the wrong thing. On the other hand, these lack of protections leads directly to its high speed.
Key point: Sometimes systems can be exploited through the cache. Examples are:
Key point: Certificates can be revoked. This means that a company who believes that their site has been compromised can put up a server on the Internet that tells everyone else that the certificate is no longer valid.
Key point: The Verisign embedded certificates in older browsers (IE 3.0, Netscape 4.0) have expiration dates of January 1, 2000. This means that anybody using older browsers will get nasty warnings when they visit ecommerce sites or attempt to verify files with authenticode.
Key point: The way it is supposed to work is that you have a certificate that claims to be "Microsoft" signed by "Verisign" (a popular CA), then you trust that Verisign has done a reasonable job both ensuring that Microsoft is who they say they are, and that Microsoft has done a reasonably good job protecting their private keys from theft.
Contrast: Microsoft could create a "self-signed" certificate, but then anybody else could create a self-signed certificate claiming to be Microsoft. Therefore, you trust a CA-signed certificate more than a self-signed certificate, as long as you trust the CA.
Key point: How do you trust a CA? The answer is marketing. First, a company like Verisign has spent millions of dollars creating a reputable company that would be destroyed if a flaw was found in their process (i.e. thieves were able to steal their private keys). Second, Versign (and a few other CAs) have managed to embed their public keys within Internet Explorer and Netscape Navigator. This means that any website using SSL must obtain a certificate signed by one of these built-in CAs, or else users get confusing warning messages.
Humor: Microsoft uses certificates signed by Verisign, because it is trusted by many people. The reason so many people trust Verisign these days is because its root keys are included with Microsoft's browsers.
Key point: One of the chief RISKS is the theft of the private key used to sign things. If a hacker/thief is able to steal it, then they can masquarade as someone
Key point: Several important CA certificates (i.e. Verisign) expired on Dec. 31, 1999. Since it is feasable to eventually compromise a certificates, they usually expire at some date. The certificates for trusting root CAs that are built-in many browsers (Internet Explorer 4.0 and earlier, Netscape Navigator 4.06 and earlier) were created in 1995, and were made for a 5-year lifespan. One of the creators of these certificates now says he wished he'd put the expiration date a little off, such as on Dec. 15, in order to avoid the Y2K madness.
Key point: The word "CGI" stands for "Common Gateway Interface", which generally confuses people more than help them.
Key point: In most cases the user is prompted for the password, which the client then stores in memory. In the use of smart cards, however, the system may give the user the challenge string, which the user then types into the smart card. The smart card then produces a response, which the user must type back into the system. In this way, the user validates that they have the smart card.
Key point: Challenge-response systems are thought to be more secure because the challenge/response is different every time. This guards against replay attacks as well as making cracking more difficult.
Key point: Favorite because it provides real-time anonymous communication.
Key point: Checksums are not secure against intentional changes by hackers. For that, you need a cryptographic hash.
Key point: A block cipher is one that encrypts a block of data at a time. For example, DES uses a block size of 64-bits. Each input block must correspond to exactly one output block (like a code-book). A block-cipher suffers from the fact the same data repeated in a message would be encoded in the same way. Consider a block size of 8-bit encrypting English text; you could therefore figure out all the letter 'e's in the cipher text because they are the most common letter used. Therefore, block-ciphers are often used in a chaining mode such that the same pattern will indeed be decrypted differently.
Contrast: clear-text, plaintext.
Misconception: The word "text" comes from traditional cryptography that meant the text of messages, though these days "text" can refer to binary computer data as well.
Key point: In block-ciphers, the key represents a code-book. In other words, you could use the key to generate a huge book of matching pairs whereby each plaintext block would match to exactly one ciphertext block. Then, you could encrypt messages by looking them up in this table.
Key point: The term ECB or Electronic Code-Book refers to the use this mode of using a block-cipher. However, since it leaks information, many people prefer to chain blocks of ciphertext and plaintext together in order to make sure that the same pattern will be encrypted differently when it appears multiple times in a message.
Key point: Since encrypted data is essentially random data, you cannot compress it. This defeats networking standards designed to automatically encrypt traffic (such as modems). Therefore, data must be compressed before it is encrypted. For this reason, compression is becoming an automatic feature to most encryption products. The most often used compression standard is gzip and its compression library zlib.
Key point: Cookies are not a security hole in themselves. However, they can be combined in interested ways with other browser features in order to create big security and privacy holes.
Key point: Turning off cookies is not practical. The best you can hope for is "cookie management" -- choose which sites you want to allow cookies for but deny them to all the rest.
Key point: One rootkit uses ICMP as a covert channel. It creates a virtual TCP-like circuit inside of ping packets.
Key point: Covert channels can become extremely covert. In theory, one can create a covert channel where only the IP identification field (16-bits) carries the data.
Key point: URLs and DNS queries pass through virtually everything (including proxies). Therefore, it is easy to export information from inside a company to the outside using this technique.
History: When the UNIX operating system was first developed, passwords were stored in the file /etc/passwd. This file was made readable by everyone, but the passwords were encrypted so that a user could not figure out who a person's password was. The passwords were encrypted in such a manner that you could test a password to see if it was valid, but you really couldn't decrypt the entry. (Note: not even administrators are able to figure out user's passwords; they can change them, but not decrypt them). However, a program called "crack" was developed that would simply test all the words in the dictionary against the passwords in /etc/passwd. This would find all user accounts whose passwords where chosen from the dictionary. Typical dictionaries also included people's names since a common practice is to choose a spouse's or child's name.
Contrast: A "crack" program is one that takes existing encrypted passwords and attempts to find some that are "weak" and easily discovered. However, it is not a "password guessing" program that tries to login with many passwords, that is known as a grind
Key point: The sources of encrypted passwords typically include the following:
Key point: The "crack" program is a useful tool for system administrators. By running the program on their own systems, they can quickly find users who have chosen weak passwords. In other words, it is a policy enforcement tool.
Tools: on UNIX, the most commonly used program is called simply "crack". On Windows, a popular program is called "l0phtCrack" from http://www.l0pht.com.
Controversy: See the word hacker for a disagreement about the way that "cracker" is used in the computer enthusiast community vs. the security community.
Culture: Cracking programs is its own little underground 'scene' independent of other hacking activities. Groups and individuals often compete to be the first to break a new copy protection scheme in popular programs. There are many sites that catalogue cracked programs.
Key point: The different kinds of cryptanalysis are:
History: So far, there are four major eras in cryptography.
The best example is the "checksum" vs. "hash".
A checksum verifies that data hasn't been corrupted unintentionally. For example, all IP packets are checksumed in case they corrupted accidentally between sender and receiver.
A cryptographic hash verifies that data hasn't been corrupted intentionally. Hackers can (and do) alter IP packets between the sender and receiver in order to carry out an attack. Since IP's checksum is not cryptographically secure against hackers.
There are two features that are required in order to be cryptographic. The first is that the algorithm be secure against attack. A checksum uses simple addtion, while hashes use a complex set of mathematical operations. The second is that the key must be of a sufficient size in order to prevent brute force attacks. The IP checksum is only two-bytes long, so that even if the algorithm were secure, it would require only 65536 tries for the hacker to get it right, which can be done in real-time.
Key point: There are sites, like http://www.attrition.org that catalogue defaced sites and mirror the defaced web-pages.
Key point: Defaced web-pages is an important part of hacker attitude.
Key point: Elite hackers rarely deface web-pages, they instead break in and control the server for other nefarious purposes that yield more profit.
Key point: Web servers are easy to deface because the average OS and web server contains vulnerabilities (defaults and samples) upon installation. It takes extensive effort to harden a server.
Key point: Security irritates customers who prefer products that are easy to use. Therefore, most vendors make the same trade off. They ship their systems with the best "out-of-box" experience, and as a result most boxes are easily hacked in their default state. The more a vendor touts its ease-of-use, the more likely hackers will find that vendor's products easy to hack.
See also: samples
Example: The Ping of Death exploit crashed most machines vintage 1995 by sending illegally fragmented packets at a vicitm.
Culture: A common word for DoS is "nuke", which was first popularized by the WinNuke program (a simple ping-of-death expoit script. These days, "nukes" are thos DoS exploits that script kiddies in chat rooms use against each other.
Key point: It takes only a couple minutes to run through hundreds of thousands of words in a dictionary in order to crack a password. Therefore, never choose a word that might be in a dictionary.
Example: Many programs contain built-in HTTP servers. This allows the program to be remotely managed from any web browser. These servers expect that only the files in their own directory and below will be read. However, hackers can still provide URLs that go up directories, and down into other directories in order to read any file from the system. For example, a hacker might be able to read the UNIX password file by typing in the URL: http://www.robertgraham.com/../../../etc/passwd.
Key point: This bug occurs because programmers frequently forget to doublecheck input.
Example: This bug is common. The original version of Win95 had this bug, so that if you had access to File and Print Sharing to any subdirectory, you also had access to the entire system. A huge number of HTTP servers and CGI scripts have this bug. Many FTP servers have had this bug.
Key point: Win9x has the quirk that three dots "..." means "two directories up", four dots "...." means "three directories up", and so on. Additionally, whereas on many UNIX systems going up past the top directory automatically generates an error, going above the top directory on Windows leaves you in the top directory. Therefore, filenames like "............/Windows/greg.pwl" are frequently seen: the hacker puts more than enough dots in the path in order to guarantee they reach the root directory.
Key point: Many popular Windows "personal web servers", including several versions shipped from Microsoft, have had either the "../.." or "....." vulnerability. In particular, since the "....." issue is not widely know, it is very common among those products that fix the first variant. FrontPage98 from Microsoft shipped with this bug.
History: DNS is relatively new. When the Internet was small, every machine simply had a list of all other machines on the Internet (stored in /etc/hosts). Generally, people just had the IP addresses of machines memorized in much the same way that people memorize phone numbers today.
Key point: DNS is not needed for communication. If a DNS server goes down, newbies will think that the entire network is down. Hackers frequently deal with raw IP addresses, and indeed often bypass DNS entirely as it may give off signs of an attack.
Key point: DNS is a "directory service".
Key point: The DNS hierarchy starts from the "top level domains" of .com, .net, .org, .edu, .giv, .mil, and the two-letter country codes (e.g. .us for United States, .jp for Japan).
Misconception: Both IP addresses and domain names use dots: "www.robertgraham.com" vs. "192.0.2.133". This has no significance; the usage of these dots is basically unrelated. Trying to match things up one-to-one is wrong (i.e. ".com" == "192.").
Misconception: Names and addresses are completely independent. Many people think that *.robertgraham.com might map to all the addresses under 192.0.2.*.
Analogy: What is your phone number? If I asked you this, you might give me both your home number and your cell phone number. I can reach you at either one. In much the same way, the a domain name like www.yahoo.com can have multiple IP addresses. Every time you visit that site, you might go to a separate IP address. You can test this out yourself. Go to the command-line and type "ping www.yahoo.com". Notice how it comes back with an IP address that it pings. After that runs, try it again. Notice how the second time it is pinging a different IP address.
Key point: Dumpster diving is generally legal, as long as you are not trespassing.
Culture: Wasn't really part of the hacker vernacular until the 1995 movie Hackers.
Culture: This word finds itself mangled in many variations: leet, 1337, 31337, etc.
Statistics: Ira Winkler, former analyst at the NSA and now write, estimates that as of 1999, that there are roughly 500 to 1000 "elite" hackers capable of finding new security holes, and roughly 5000 hackers capable of creating exploit scripts. (He further estimates about 100,000 script kiddies).
Key point: Encryption has massive philosophical implications. Widespread use of encryption means that people can hide their data from governments (especially repressive ones) and law enforcement (especially when you are committing a crime).
Contrast: Asymmetric encryption uses different keys for encryption and decryption. Since the most useful form of this is one you keep one key private and make the other public, this is better known as public key encryption. In contrast, symmetric encryption uses the same key for both encryption and decryption.
Example: Some algorithms popular in cryptography are: DES, rc4
Example: Some popular applications that use encryption are: PGP
Example: Some protocols that use encryption are: SSL, IPsec
Analogy: For example, one company provide software that another company sells imbedded in their hardware. The second company (the OEM) is scared that the first company might go out of business, so requests that the first company put the source code for the software in escrow. Should the first company go out of business, the second company would still be able to sell their product.
Key point: Law enforcement is constantly pushing for key escrow where a third party holds back-door keys to all encryption products. Law enforcement would then be able to obtain these keys with a court order into order to decrypt messages or eavesdrop on communications. They first propose a variant of the two-person rule in order to prevent abuse of the system. Note that the general problem is called key recovery (in which law enforcement can recover the key using some means); key escrow is just one way of doing key recovery.
Key point: The need to exchange keys is the reason encryption protocols are not secure. There is an absolutely secure encryptiong method called a one-time-pad. However, in practice, you cannot exchange vast quantities of one-time-pads.
Key point: PKI essentially solves the key exchange problem.
Culture: Exploits, or "exploitz", are the root of the hacker culture. Hackers gain fame by discovering an exploit. Others gain fame by writing scripts for it. Legions of script-kiddies apply the exploit to millions of systems, whether it makes sense or not.
Controversy: There is no good definition for this word. It is debated a lot trying to define exactly what is, and is not, an exploit.
Key point: Since people make the same mistakes over-and-over, exploits for very different systems start to look very much like each other. Most exploits can be classified under major categories: buffer overflow, directory traversal, defaults, samples, Denial of Service
Key point: The problem is that TCP/IP knows no boundaries. When a user tells the system to share files with the rest of the familly, the user is not quite aware that this means the files are shared with the rest of the Internet. This means that anybody, anywhere on the Internet can at any time connect to the machine and read/write files. To see if somebody has accidentally shared their hard-disk, right-hand-mouse-click on "Network Neighborhood" in Windows, select "Find Computer...", then type in that user's IP address.
Key point: File and Print Sharing used the SMB protocol over NetBIOS on TCP port 139.
Example: The following shows the output of the command "finger email@example.com":
Login: rob Name: Robert David Graham Directory: /home/rob Shell: /bin/bash On since Fri Dec 3 18:13 (PST) on ttyp0 from gemini No mail. No Plan
Key point: The finger command reveals extensive information. For example, if I were attacking the above machine, I would notice that the user is running bash Therefore, I might try something like http://rh5.robertgraham.com/~rob/.bash_history against the user, which in about 1% of the cases will give me a history file of recent commands they've entered, which might contain passwords and such.
A firewall acts as a "choke point". Corporations setup firewalls between their internal networks and the Internet. All traffic between the corporation and the Internet flows through the firewall. It acts as a "gate" with virtual guards that examines the traffic, and decided whether to allow it or block it.
Misconception: Many people believe that a firewall makes your network immune to hacker penetration. Firewalls have no ability to decide for themselves whether traffic is hostile or benign. Instead, the administrator must program the firewall with rules as to what type of traffic to allow or deny. This is similar to a guard checking badges at a gate: the guard can only detect if the badge is allowed/denied, but cannot detect impersonations or somebody climbing the fence in the back.
Key point: Firewalls are based on the principle of blocking everything by default and only allowing those things that are absolutely necessary.
Key point: Firewall administrators are frequently at odds with their management. Executives are frequently frustrated by things that don't work in the network. They don't understand how difficult it is to secure each new application, or the increased risks involved.
Controversy: A lot of time is wasted on trying to come up with the exact definition of the word "firewall", usually by marketing flaks or nerds with attitude. The term isn't well defined. In my definition above, I chose the language explicitly to hilite the fact that firewalls do not have the ability to determine "right from wrong" (like an intrusion detection system), but instead simply closes off most Internet access in the hopes that it closes unknown holes in the network.
Misunderstanding: A common question posed is "what is the best firewall?". This question comes from the belief that a firewall is a magical device that you plug into your network in order to stop hackers. Instead, a firewall is a device that isolates your network from the rest of the Internet.
Example: Ethernet supports a maximum packet size of 1500 bytes. Therefore, in order to send an IP packet of 2000 bytes, the system must first fragment the packet into two pieces before transmission. The other end will then reassemble them back into a single packet on the other end.
Contrast: The general concept of fragmentation applies to all layers of the protocol stack. For example, ATM has a maximum frame size of 48-bytes, which is too small and inefficient for any purpose if higher layers had to deal with it. Therefore, the ATM adapter itself handles the fragmentation and presents a "virtual" interface that allows a full 64-kilobyte packet to be sent without IP level fragmentation. Conversely, when reading files from a file server, even a 64-kilobyte packet size is too small, so the file server layer automatically requests smaller parts of the file. In some cases, applications will attempt to calculate the MTU (Maximum Transmission Unit) of the connection in order to optimize operations to avoid any IP fragmentation.
Key point: IP fragmentation is slow, and is better handled either below the IP layer (like ATM) or above it (like in the application layer).
Key point: Fragmentation and reassembly is difficult to program right. Therefore, there are numerous ways to hack this feature. Some attacks are:
Key point: Most network-based intrusion detection systems do not reassemble packets. Therefore, a hacker can use something like fragrouter in order to evade the IDS.
Key point: Fragmentation is almost never needed. Most communication runs over TCP, which does its own "segmentation" which is more efficient. Therefore, if you see any fragmentation on your network, you should examine it closely to see if it indicates an attack.
Key point: FTP uses an outgoing control connection that only sends commands to the server and receives returned status information. All data is transfered on separate connections (one connection for each file or directory transfered).
Key point: The control connection is text based, so you can use Telnet or netcat as your client (if you understand the protocol).
Key point: Before the web (and graphical browsers) people used command-line versions of FTP. These are still prefered by hackers, becauase GUIs are often too "noisy" (generating unnecessary commands). Such command-line clients that are still included in virtually all UNIX or Windows systems.
Key point: These separate connections are created by sending a PORT command across the control connection. This command accepts both and IP address as well as port number that tells the other side where to connect. Example: PORT 192,2,0,201,10,1 is the string sent across the control connection to tell the server that the client has opened a port on the machine with the IP address 188.8.131.52 with port 2561. The server will then open up a TCP connection as intructed. This command is sent invisibly when the client requests a directory listing or file; all the client sees of this happening is a status message to the effect 200 PORT command successful. which is sent back across the control connection. A neat hack is to specify somebody else's IP address in this command. This hack is called a bounce attack, and can be used to port scan computers or subvert trust relationships.
Key point: An outgoing connection is used for control, but the data is sent on an incoming connection. Firewalls block incoming connections. Therefore, a user will see that they can connect to the FTP server, but directory listings and file transfers don't work.
Key point: In order to solve the incoming connection problem, FTP supports a mode called PASV that forces all connections to be outgoing. Web-browsers like IE and Netscape use PASV mode by default. Command-line FTP clients typically don't support PASV; but people try "quote PASV" commands anyway.
Key point: Lots of FTP servers have buffer overflow exploits in them.
Analogy: If someone steals your bank card, they cannot sit in front of the cash machine and guess all possible PIN numbers. After a certain number of unsuccessful tries, the bank machine will "eat" the card.
Key point: Secure systems (UNIX, Windows NT) lock out accounts after a certain number of unsuccessful tries. These lock-outs can either be temporary (and restore themselves automatically), or permanent until an administrator intervene and unlocks the account.
Key point: Non-secure systems (Win9x and many software applications) do not lock out accounts. For example, if you have Win9x "File and Print Sharing" turned on and protected with a password, a hacker can try continuously and invisibly to gain access to your machine. Nothing is logged, nothing is locked out.
History: The word comes from the book Stranger in a Strange Land by Robert Heinlein. This was a popular counter-culture book in the 1960s, and is a popular Science Fiction book today.
Key point: One of the precepts of Zen philosophy is that the important concepts of life cannot be described by words, and therefore there exists no written description to the path of enlightenment. Grokking means to understand something at a level beyond what mere words can express.
Key point: There are three levels of understanding, which can be illustrated by looking at a cars engine. At the first level, people look at all the parts and say to themselves "This is unnecessarily complicated, I'm sure there is a way we can remove many of these parts and make it simpler". Probably 99% of the population approaches life in this manner. The second level is an engineer who understands how the engine works, and how the various parts work together in the ingeneous fashion that they do. This engineer understands that this the simplest way to produce an engine, and that it has reached this stage after years of being perfected by countless engineers. At the third level is the godlike engineer that understands how to remove one part in order to make the engine even simpler. In this analogy, the engine is the computer. Likewise, the Internet is populated by script-kiddies who are constantly searching for ways to learn about hacking without being bothered by all the unnecessary complexity.
Key point: The failure to grok is often due to failure to understand the correct abstractions. Understanding a thing requires understanding the context in which that thing lives. If one cannot step out of a traditional context in order to regard a thing within the proper context, one cannot grok it. For example, many people have trouble grok the layering of network protocol because the only can only see what the protocols due for them, not what the protocols due in general. Therefore, when they look at protocols, all they see is large amounts of inscrutable unnecessary complexity.
Consider Arthur C. Clark's Third Law: "Any sufficiently advanced technology is indistinguishable from magic". Since normal people have no clue as to how computers work, they often view hackers with suspicion and awe (as magicians, sorcerers, witches, and warlocks). This suspicion leads to the word "hacker" having the connotation of someone up to no good.
History: The word "hacker" started out in the 14th century to mean somebody who was inexperienced or unskilled at a particular activity (such as a golf hacker).
In the 1970s, the word "hacker" was used by computer enthusiasts to refer to themselves. This reflected the way enthusiasts approach computers: they eschew formal education and play around with the computer until they can get it to work. (In much the same way, a golf hacker keeps hacking at the golf ball until they get it in the hole).
Furthermore, as "experts" learn about the technology, the more they realize how much they don't know (especially about the implications of technology). When experts refer to themselves as "hackers", they are making a Socratic statement that they truely know nothing. For more information on this connotation, see There is still a large community of enthusiasts who use this word with only that connotation, such as in ESR's "Hacker Dictionary".
Key point: Today if you do a quick search of "hacker" in a search engine, you will still occasional uses of the word in senses used in the 1400s and 1970s, but the overwhelming usage in the 1990s describes people who break into computers using their sorcerous ways. Likewise, the vast majority of websites with the word "hack" in their title refer to illegitimate entry into computer systems, with notable exceptions like http://www.hacker.com (which refers to golf).
Controversy: The computer-enthusiast community often refers to any malicious hacker as a "cracker". The security-community restricts the use of the word "cracker" to some who breaks encryption and copy-protection schemes.
Consequently, a journalist who writes about cybercriminals cannot use either word without hate mail from the opposing community claiming they are using the word incorrectly. If a journalists writes about hackers breaking into computers, they will receive hate-mail claiming that not all hackers are malicious, and the that the correct word is "cracker". Likewise, if they write about crackers breaking into computers, they will receive hate-mail claiming that crackes only break codes, but its hackers who break into systems. The best choice probably depends upon the audience; for example one should definately talk about malicious crackers in a computer-enthusiast magazine like "Linux Today".
Example: The program "tripwire" detects intrusions by calculating a hash of all programs. On a regular basis, it recalculates the hash. If a file has changed, then tripwire will detect a change in the hash. Therefore, one of the first things hackers will do when breaking into a system is to search for such processes running. (Simply looking for the md5 program is a dead giveaway).
Key point: A hash is "one-way" or "nonreversable". This means that a hash cannot be used to recover the original data.
Key point: A typical hash creates a 128-bit value. This means that there must exist multiple messages that generate the same hash. However, while this can happen in theory, we pretend it can't happen in practice. This is less likely to happen than an asteroid colliding with the Earth destroying all life within the next 100 years.
Key point: Another word for "hash" is message digest.
Key point: The MD5 (Message Digest 5) is the most popular hashing algorithm at this point.
Example: The following quote describes a social engineering attack:
The Hitchhiker's Guide to the Galaxy has a few things to say on the subject of towels.
A towel, it says, is about the most massively useful thing an interstellar hitchhiker can have. Partly it has great practical value. You can wrap it around you for warmth as you bound across the cold moons of Jaglan Beta; you can lie on it on the brilliant marble-sanded beaches of Santraginus V, inhaling the heady sea vapors; you can sleep under it beneath the stars which shine so redly on the desert world of Kakrafoon; use it to sail a miniraft down the slow heavy River Moth; wet it for use in hand-to-hand combat; wrap it round your head to ward off noxious fumes or avoid the gaze of the Ravenous Bugblatter Beast of Traal (a mind-bogglingly stupid animal, it assumes that if you can't see it, it can't see you-daft as a brush, but very very ravenous); you can wave your towel in emergencies as a distress signal, and of course dry yourself off with it if it still seems to be clean enough.
More importantly, a towel has immense psychological value. For some reason, if a strag (strag: nonhitchhiker) discovers that a hitchhiker has his towel with him, he will automatically assume that he is also in possession of a toothbrush, washcloth, soap, tin of biscuits, flask, compass, map, ball of string, gnat spray, wet-weather gear, space suit, etc., etc. Furthermore, the strag will then happily lend the hitchhiker any of these or a dozen other items that the hitchhiker might accidentally have "lost." What the strag will think is that any man who can hitch the length and breadth of the Galaxy, rough it, slum it, struggle against terrible odds, win through and still know where his towel is, is clearly a man to be reckoned with.
Hence a phrase that has passed into hitchhiking slang, as in "Hey, you sass that hoopy Ford Prefect? There's a frood who really knows where his towel is." (Sass: know, be aware of, meet, have sex with; hoopy: really together guy; frood: really amazingly together guy.)
Key point: The answer to life, the universe, and everything is 42.
Example: ISPs generally reassign IP addresses of dialing users very quickly after a previous user hung up. Take for example where Alice dials up the Internet, telnets to a host, then for some reason hangs up without gracefully closing the connection. Now consider Mark, who dials-up later and is assigned the same IP address. Let's say that Mark has created his own TCP/IP stack that automatically hijacks any existing connection. The server then sends some response packet back across the connection to Alice (really Mark). At that point, Mark's stack automatically picks up the connection and continues the protocol. At this point, Mark can do anything he wants on Alice's account.
Example: Similar to above, hackers often hijack connections by first nuking one end of the connection, then spoofing that side's IP address.
Example: Spammers scour the Internet looking for open USENET NNTP servers. If they find a server they can post floods of spam through, this is known as "hijacking" the server.
Misconception: A common misconception is that by advertising the system or inviting hackers in causes you to lose all rights to prosecute the hacker. Honeypots do not advertise themselves nor invite hackers. They simply sit on the network waiting to be discovered and hacked. If a hacker doesn't search them out, they won't find them. Similarly, honeypots can contain legal notices in their banners telling hackers to go away.
Culture: The term is an outgrowth of the older abbreviation "h/p" (hack/phreak).
Key point: HTTP is text based, so you can use Telnet or netcat as your client (if you understand the protocol). For example, you can telnet www.example.com 80 to connect to a web-service and enter the command GET / HTTP/1.0<cr><cr> in order to download the home page.
Misconception: Packet filtering firewalls work by filtering source/destination ports in the TCP and UDP transport protocols. However, as a secondary function, they also filter ICMP type and code numbers. They frequently map these to where UDP and TCP have port fields in both the configuration step and the output logs. These concepts are unrelated, however.
Contrast: A host-based IDS monitor system events, logfiles, and so forth. A network-based IDS monitors network traffic, usually promiscuously.
Key point: At the end of 1999, all freshly installed IIS v4.0 servers were vulnerable to the .htr buffer overflow bug and the RDO exploit. Roughly 90% of IIS servers are not sufficiently hardened against these exploits, and are thus vulnerable to being owned or defaced.
Key point: IMAP is important to hackers because many implementations are vulnerable to buffer overflow exploits. In particular, a popular distribution of Linux shipped with a vulnerable IMAP service that was enabled by default. Therefore, even today, security professionals frequently detect scans directed at port 143 looking for vulnerable IMAP servers.
Key point: The following are useful resources to such a team:
The file /etc/inetd.conf configures this service.
See also: buffer overflow, directory traversal
Key point: All data on the Internet is carried by IP packets.
Key point: IP is an unreliable datagram protocol, meaning that routers may sometimes drop packets during congestion. A protocol like TCP must be added to IP in order to track packets and resend them if necessary.
Key point: The ability to manipulate IP headers by programs is limited, so there are few defenses against such techniques. Many hacks rely upon low-level manipulation of headers.
Key point: The IP header is shown below. Since IP is carried across a link between router-router or host-router, link headers like Ethernet, PPP, etc. may come before this header. Likewise, the payload of the IP packet comes after this header.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |Version| IHL |Type of Service| Total Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Identification |Flags| Fragment Offset | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Time to Live | Protocol | Header Checksum | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Source Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Options | Padding | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|This 4-bit field always has a value of "0100" (binary) or "4" decimal. Many plan to replace IPv4 with the much more complex IPv6 in order to solve addressing and security issues.|
(Initial Header Length)
|Indicates the length of the IP header. The length of the header is always "20-bytes" unless options are present.|
|Type of Service |
|Not really used, the ToS field gives hints to the router how the packet should be routed. The typical example is a connection between Las Angeles and New York where a router can choose to send the packet across a low-speed land-line (dial-up) vs. a high-speed satellite connection. The latency for a land-line is a few milliseconds, whereas a satellite can be about a second. Therefore, you want the low-latency for interactive connections like Telnet, but you want the high badwidth for connections like FTP. Since this field isn't really used that much, hackers can use it as a covert channel.|
The total length of the IP datagram once the packet has been reassembled.
A unique ID number for the entire packet. All fragments of a packet
carry the same ID.
Key point: Tracking the ID field over time can help fingerpint the OS.
Key point: Some systems use monotonically increasing IDs, so you monitor activity on a remote machine by pinging it on a regular basis.
Key point: A covert channel can be created by encapsulating information in this field.
There are two flags that control fragmentation.
The DF (Don't Fragment) bit
tells routers not to fragment this packet. The MF (More Fragments) bit
indicates that this is not the last fragment in the packet.
Key point: You can evade network-based IDS sometimes by careful use of the DF bit and oversized packets that must be fragmented.
The offset from the start of the original packet that this fragment starts.
Key point: The Ping-of-Death exploit resulted by combining a fragment offset plus fragment length in order to exceed the maximum IP packet size.
|Time to Live |
This field indicates how many hops (routers) the packet can pass through
before being discarded. Each router who forwards the packet decrements
this field by one. When a router decrements the field to zero, it
assumes a routing loop has occurred and sends back an ICMP
message back to the sender.
Key point: Abuse of the TTL field, after fragmentation is the most useful technique for manipulating IP headers. In addition, it is easy to manipulate this field at the sockets layer.
Key point: The traceroute program finds all the routers in the path to a target by sending out many packets with varying TTL fields. This causes every router to receive a TTL in one of the packets that it zeroes out, causing it to report its existance back to traceroute.
Key point: Tracerouting through firewalls is sometimes possible by adjusting the TTL of TCP replies.
This field indicates the next protocol header after the IP header. Examples
are a value of 1 for ICMP, 6 for TCP,
and 17 for UDP.
Key point: Some rootkits use this as a way of invisibly transporting data since most systems cannot detect or log unknown protocols at this layer.
The IP address of who sent the packet. This is included
in every packet so that the destination knows who to respond to, and any
errors can likewise be sent back to the sender.
Key point: The IP address can be forged (spoofed). This can sometimes be useful despite the fact that it causes any responses to be sent back to the spoofed IP address rather than the real sender.
|The IP address of where the packet is going to. Each router along the way compares this IP address to internal routing tables in order to figure out which direction to forward the packet.|
Additional options that can affect how the packet is routed. Multiple options
can be specified.
Key point: 99.999% of all IP packets have no options. Some IDSs trigger simply whenever they see an option field.
Key point: The most common option used for attacks is source routing.
|An IP headers must be aligned on even 32-bit boundaries, which may sometimes require nul bytes to be added.|
Key point: The IP address shows up inadvertently in many communications. By examing the details of e-mail headers, you can usually find the IP address that somebody sent e-mail from -- even if the user is behind a firewall. This a common way that sui dissant hackers are caught: they attempt to use anonymous e-mail services to send mail, only to be caught by the inclusion of their IP address in the headers.
Key point: IPsec's main use today is when tunneling traffic for VPNs. It can also work for generic encryption of data between two hosts.
History: This was the name for the U.S. military campaign during WW-II in order to take over islands closer and closer to the enemy, using each new island as a base from which to launch further attacks.
Key point: University systems, which are based upon the idea of openess and free sharing, are a hot-bed of compromised systems from which hackers launch attacks. Increasingly, home user machines attached to DSL lines and cable-modems are being compromised and used to launch attacks from.
Contrast: People are confused as to the difference between a key and a password. The obvious difference is that a password is a variable number of characters typed in by the user, whereas a key is a fixed-length binary value that that the user never sees. This begs the question: how do I decrypt the file with my key if it is a binary value I never see? The first answer is that a password is often converted into a key by taking a hash of the password. In more complex (and more common) cases, please read the section on "public keys"
Contrast: There are two types of keys: symmetric and asymmetric (aka public/private key).
Contrast: A common question deals with the difference between 40-bit
and 128-bit encryption in web browsers like Netscape. The answer is that
the most obvious way to break the encryption and read the plain text
is to simply try all possible keys. A 40-bit key has
roughly one trillion (
Key point: Moore's Law breaks all cryptosystems, eventually. This, and only this, is why DES has become obsolete. Note: 40-bit and 128-bit keys refers to the RC4 algorithm used within web-browsers to talk to web servers via SSL. The U.S. restricts export of all software whose keys are greater than 40-bits in order to be able to spy on foreigners (ostensibly only in a military engagement).
Key point: Keys are generally just "session keys". This means they are dynamically generated at the beginning of a session, and exchanged with the partner using PKI
Key point: Once keystrokes are logged, they are shipped raw to the hacker. The hacker then peruses them carefully in the hopes of either finding passwords, or possibly other useful information that could be used to compromise the system or be used in a social engineering attack. For example, a key logger will reveal the contents of all e-mail composed by the user.
Key point: Keylog programs are commonly included in rootkits and remote access trojans.
Key point: You can also purchase hardware devices that plugin between the keyboard and the main system (for PCs). These are OS independent, they simply start recording, then the hacker can retrieve the device and instruct it to simply spit out all the characters back again on the hackers system.
Contrast: Without known plaintext, key cracking is only a little bit more difficult. Running heuristics on the output of the decryption engine makes the decryption several times harder, but when you think about this, it only means that making it four times harder is only equivelent to adding 2 bits to the key length.
Key point: Most messages contain "headers" that represent known-plaintext. IP packets all have similar headers. E-mail message all contain the same fields. Therefore, there is a fair amount of implicit plaintext that is known even when a decrypted sample of the message doesn't exist.
Example: System administrators typically have multiple accounts with different rights. For example, when I'm logged in as a normal user, I do not have rights to administor my own machine. I must login as a separeate account in order to do such things, then log out as soon as I'm done.
Key point: This often means that both sides of a connection really need to authenticate themselves. For example, when you log into a server, you really want to be assured it is the real server you are talking to, rather than Mark who is forwarding your requests to the real server using your identity.
The idea is that the power of the Internet is not simply all the websites that you can access (linear), but the power represented by everyone else also on the Internet (exponential). For example, organizations like http://www.distributed.net cannot only harnass lots of machines in order to tackle large problems (linear), but they also can exploit the word-of-mouth on the Internet to sign up (exponential). Similarly, consider the growth in sites like http://www.slashdot.org that start out as hobbiest sites, but eventually blossom into large money making ventures, tossing pre-Internet-age business philosophies on their ear.
Key point: Hacker attacks grow exponentially because more and more hackers are getting online (especially from 3rd world countries) and more and more resources (businesses) are getting online.
Key point: The amount of computing resources a hacker can tap into from his/her computer desktop is more than the combined might of all governments and militaries.
Key point: A big problem with corporations is that they do not spend enough time hardening mission critical applications, or spend too much effort on non-mission critical elements.
Key point: Every bit of key length doubles the security, making it twice as difficult to crack. However, because of Moore's Law, every year that passes makes all keys twice as easy to crack. Therefore, if it takes 1-week to break a message encrypted with a 40-bit key, it will likewise take 1-week to break a message encrypted with a 128-bit key roughly 100 years from now.
Contrast: A NAT provides some firewalling capabilities because isolates the end-nodes while still providing access to the Internet. The isolation is better than packet-filter firewalls, but not as good as proxies.
Misconception: Like sockets, many different protocols can be used to transport applications written to the NetBIOS API. When you say "NetBIOS", some people will understand you to mean the TCP/IP transport. Other people will think of "NetBEUI", which is the transport over raw Ethernet without any intervening routable network protocol. Use the term "NBT" (NetBIOS-over-TCP) or "NetBEUI" to avoid confusion.
Contrast: Microsoft's "File and Print Sharing" uses the SMB protocol over NetBIOS. Microsoft supports the NetBIOS interface over TCP/IP, NetBEUI, and Novell's IPX/SPX. Home users who share files among their own machines mistakenly enable File and Print Sharing using the TCP/IP transport, allowing hackers anywhere on the Internet access to their machine. Instead, they should configure it over the NetBEUI transport so that nobody outside their network can access their files (note: this still might open up their networks to people on the same cable-modem VLAN).
Key point: Hardened WinNT computers should use NTFS exclusively. After installation, some file and directory permissions need to be adjusted.
Key point: NTFS supports a feature called "alternate datastreams" (similar to Macintosh data and resource forks) that can be used to attach data to files in a hidden way. The additional information does not appear to change the size of the file in directory listings.
Example: Many people put their POP3 service at a different port than port 110 as a way to foil attacks against that port. However, a simple port scan reveals this port, and the banner reveals what service is running at that port.
Key point: The security landscape is littered with foolish people who try short-cuts around the tried-and-true techniques.
Key point: Obscuring information always helps security. The only question is whether it is your prime security defense, or an adjunct to it.
Key point: Do the math: if you obscure things such that a successful attack is 99% less likely, this means 10 out of a thousand attacks will succeed. In general, hackers have the resources to through billions of attacks against an obstacle: it's just automated computer time and network traffic.
Key point: If one-time pads are unbreakable, then why aren't they used? The major problem is how one distributes the one-time pads to all the receivers. This can be done in some cases, such as sending out CD-ROMs full of random bits with soldiers on the battle-fields, but it becomes unwieldy for normal uses of cryptography.
Key point: OSPF can easily be subverted by "local" hackers. This means that hacker generally has to be within network area he/she wishes to subvert. The hacker will either pretend to be a router, or spoof packets from nearby routers. The most common technique is to "advertise" false information (spoof Link State Advertisements (LSA)).
Culture: To control completely. A machine broken into and under completely under control of the hacker is "owned".
See also: attitude
Analogy: Imagine looking at an automobile freeway during rush hour from an airplane. The freeway looks like a flowing river, but each individual car (packet) is really independent from all the others. While it looks like the cars on the freeway are going in the same direction, each car really has its own source and destination separate from the others around it. This is how Internet core routes look.
Analogy: Now consider that a bunch of coworkers leave the office and go to a party. Each gets in his/her own car and drives to the party. Each person may take a slightly route, but they all end up together at the party. This demonstrates how data is broken up into individual packets, sent across the Internet (potentially following different routes), then reassembled back again at the destination.
Key point: Conceptually, networking occurs at abstract layers well above the concept of packets. Users type in a URL, and the file is downloaded. By dealing with the raw packets themselves, hackers are frequently able to subvert communications in ways not detectable at these higher layers.
Contrast: The term "packet switched network (PSN)" is used to describe the Internet, whereas the term "circuit switched network (CSN)" is used to contrast it with the traditional phone system. The key difference is that in the phone system, the route between two people is setup at the start, and each bit in the stream follows that route. On the Internet, each packet finds its own route through the system, so during a conversation, the packets can follow different paths, and indeed arrive out-of-order. Another key difference is latency. The phone system forwards each bit one at a time, so as soon as one arrives, it doesn't have to wait before forwarding it on. On the Internet, bits are bunched together before transmission. Each hop must wait and receive all the bits before forwarding any of them on. Each hop therefore adds a significant amount of delay. Gamers know this as the "ping" time.
Key point: There are other technologies that use packets, not just the Internet. Before the Internet came along, X.25 networks were a popular form of packet-based communication (and indeed, X.25 formed the basis for many links on the nascent Internet).
Key point: The most important defensive mechanism that a corporation can take is to create and enforce policies about proper password usage.
Key point: A leading cause of compromise are programs that leave behind default passwords. A leading cause of compromise are users who choose weak passwords that can easily be guessed or cracked.
Tools: The crack programs can be used to maintain a strong policy>
Tools: On Windows NT, the "passflt.dll" and "passprop.exe" tools can be used to enforce strong passwords.
Misconception: People used to believe that a good password was a random mix of UPPER and lower case, numbers, and punctuation. However, this generates passwords that are impossible for users to remember, so they find ways around the restriction, such as writing passwords down on Post-It notes. Therefore, somebody can compromise the network by simply looking for Post-It notes (such as pasted to the bottom of a keyboard).
Controversy: Many policies declare that a password must be changed frequently, and most OSes come with tools for enforcing this. However, this leads to the same problem as above: it causes pain for users, so they behave in ways that reduce security. Also, it isn't clear that it dramatically increases security.
Contrast: Passwords aren't the only authentication scheme possible. Crypto-cards are often used to generate "one-time passwords" or challenge-response authentication.
Tip: Use a PalmPilot and a crypt program to store your many passwords.
See also: grind, crack, password cache, 8-character password.
Key point: All systems cache passwords in memory during a login session. Therefore, if a hacker can gain access to all memory on the system, he/she can likely sift the memory for passwords. Likewise, hackers can frequently sift pagefiles for passwords.
Key point: Many programs whose goal is ease-of-use will ask the user if they want to save the password on disk (in a file or registry. For example, the MS Outlook e-mail client has this feature to cache the POP3 passwords. Therefore, hackers have programs that will sift the file system or registry or these passwords. Some systems will store these cached passwords in clear-text, others attempt to encrypt the passwords, but usually this encryptiong mechanism ca be defeated.
Key point: v5 of PERL has the concept of "tainted" input that cannot be passed raw to the operating system without preprocessing. This is an amazingly useful feature that solves the majority of input validation problems in CGI scripts.
Key point: All true hackers use open-source versions of PGP to encrypt their data.
Key point: This attack has become "classic". Virtually all CGI scanners search for it, and the typical intrusion detection system will trigger on it.
Key point: It falls victim to the input validation problem in order to allow for directory traversal.
Key point: Most of the literature available on the net applies to phone systems that are 20 years old. These techniques rarely work on modern phone systems.
Key point: 2600 Hz is the frequency of the whistle that was provided in Captain Crunch cereal boxes that was also a signaling frequency for phone systems in the 1960s.
Example: The ping-of-death attack used IP fragmentation to crash systems. It was so named because the ping program built-in to Windows could be easily told to fragment packets this way.
Key point: Even though the ping program is simple, it can be abused. Some versions can be commanded to send packets as fast as possible, which is often done to flood networks. Most versions allow the packet size to be set to a large size, forcing fragmentation. When used with the flood above, it can overload machines since fragmentation reassembly is so slow.
Key point: Depending upon context, plaintext can refer to the contents of a message before encryption, after it has been decrypted, or even a message that is in the "clear" and not encrypted at all.
Key point: Many networking protocol protocols use plaintext passwords that can simply be sniffed off the wire.
Example: A specification stating that employees will not run scanning tools against the corporate network.
Key point: Nerds hate bureaucracy, and frequently resist clarifying the security stance. Policy statements are frequently useful, and small ones given to new employees help prevent a lot of problems before they start. They are also useful CYA: when an executive starts complaining about something not working through the firewall, it helps to pull out the Policy and explain that it isn't allowed.
Key point: Policy documents bring out the "process" bureaucrats who will spend hours debating the policy in order to avoid real work that they might be judged upon. As a result, many companies spend a lot of effort creating useless policy documents.
Key point: A good rule of thumb: only put things in the policy document that include a plan on how you enforce it. For example, if you say "users should choose strong passwords", include a plan on how to enforce them.
Key point: Virtually all local exploits are privilege escalation attacks.
Key point: The most common example of this attack is through setuid programs that have known bugs in them, often through buffer overflows or race conditions.
Key point: By manipulating the protocol raw themselves, hackers can do powerful things that are impossible in an application. For example, client applications typically limit the length of a username that can be typed in. By manipulating the protocol raw, hackers can supply any sized username they want, sometimes causing a buffer overflow exploit.
Key point: Protocols are either text-based or binary. Text-based protocols can be read directly off the wire and manipulated directly. Binary protocols require a protocol analyzer to decode them, and must be manipulated programmatically.
See also: See the section on "banners" for examples of what some protocols look like on the wire.
Key point: Encryption can happen at any layer.
|Payload||The data itself can be encrypted independent of the protocols used to transport it. For example, a typical use of PGP is to encrypt a message before sending via e-mail. All the e-mail programs and protocols are totally unaware that this has occurred.|
|Application Layer||Some applications have the ability to encrypt data automatically. For example, SMB can encrypt data as it goes across the wire|
|Transport Layer||SSL is essentially encryption at the transport layer.|
|Network Layer||IPsec provides encryption at the network layer, encrypting all the contents above IP, including the TCP and UDP headers themselves.|
Analogy: Consider talking to somebody who speaks a foreign language through a translator. You talk to the translator, who receives your statements, then regenerates something else completely to the other end. The translator serves as your proxy.
Key point: The communication terminates at the proxy. In other words, the proxy doesn't forward data so much as it tears it completely apart. For example, an HTTP proxy doesn't forward every request sent through it. Instead, it first examines if it already has the requested web page in its cache. If so, then it returns that page without sending another request to the destination server. Because proxies completely terminate the communication channel, they are considered a more secure firewall technology than packet filters, because they dramatically increase the isolation between the networks.
Key point: Protecting the "private key" from theft/disclosure is the most important thing any company can do. There is exist private keys whose value lie in the range of hundreds of millions if not billions of dollars (such as the key Verisign uses to sign certificates).
The private key is usually protected with strong encryption based upon a strong password. In paranoid cases, parts of the password are given to different people, so that more than one person must be present in order to recover the private key for use (note: redundancy is also used, if the the key is XYZ, then Alice knows XY, Bob knows YZ, and Charlene knows XZ, meaning that any two can unlock the private key).
The paranoid things you see in movies about high-security installations apply:
Private-keys are frequently stored on separate objects. The most common is the floppy disk, which can be inserted into a server when booted, but removed to a safety deposit box. Other examples include crypto-cards. (Note: when you get a certificates from a CA, they usually require that the private-key never be stored on a computer).
Servers that must use private keys must employ heavy countermeasures:
Key point: Most software-based random number generators are not cryptographicly secure. As a result, many cryptosystems have been broken by attacking the generator of session keys. Versions of Netscape and Microsoft browsers have been broken this way.
Key point: Modern computers come with hardware-based random number generators that base their seeds upon electronic noise. This solves the problem of the impossibility of pure-software random number generators of being truely random.
Key point: Software random number generators cannot be perfectly random, so they are called pseudo random number generators (PRNG). Because they cannot generate perfectly random numbers, they usually query the user for some random seed information when they startup. For example, when you install PGP, you type at the keyboard some random text. The program measures the time spacing between keystrokes, then saves that information in a file. That file then serves as the seed information for future cryptographic purposes. For the most part, such seeds are considered strong enough for cryptographic purposes because they represent more bits of information used in most keys. *
Key point: Software random number generators generally start with a single seed from which all other numbers are generated.
Key point: The output of hash or encryption algorithms produce what appears to be random data. In fact, their security depends upon this. The goal of cryptanalysis is to hunt deep within the data for patterns. As a byproduct, this also means that cryptographic algorithms can also be used as a random number generator.
Contrast: A trojan is any program with a hidden intent. A RAT is one whose hidden intent is to remotely control the machine. In particular, once the program is run and installs itself as a hidden background service, it ceases to a trojan in the classic sense and is now better thought of as a rootkit.
Example: Back Orifice, NetBus, SubSeven, Hack'a'tack
Contrast: A remote access trojan is not a virus. The general populace uses the word virus to apply to any hostile program a hacker might use. Normally, being a purist using the correct word is futile, but in this case the distinction is important. You catch viruses accidentally, and the virus rarely does anything hostile to your system. Conversely, when a hacker attempts to infect your system with a remote access trojan, the hacker is attacking you personally.
Key point: Infections by remote access Trojans on Windows machines are becoming as frequent as viruses. One common vector is through File and Print Sharing, when home users inadvertently open up their system to the rest of the world. If a hacker has access to the hard-drive, he/she can place the trojan in a location known as the "startup" folder. This will run the trojan the next time the user logs in. Another common vector is when the hacker simply e-mails the trojan to the user along with a social engineering hack that convinces the user to run it against their better judgement.
Key point: On WinNT machines, the registry can often be remotely accessed. Some portions can even be read without a password.
Key point: Many programs cache passwords in the registry, often in clear-text or only slightly obfuscated.
Key point: Trojan horses often place themselves in a "Run/RunService" registry entry in order to be automatically launched on the next reboot. Double-check these entries in order to improve the security of your system.
Analogy: Create a file on the disk called "rob.txt" containing the word "foo". Open the file in an editor and add the word "bar". However, before saving the file open it again in another editor and add the word "baz". Now save the first file, then save the second file, and exit both editors. The file now contains "foo baz", and the changes for "bar" are completely lost. Note: this may not work for you, because some editors check for this condition so that you can't make a mistake.
Key point: It is the most popular cipher used in secure web connections via SSL.
Key point: RC4 supports variable length keys, but US restrictions limit it to 40-keys when US companies export products using RC4.
Key point: RC4 was a trade secret until somebody reverse engineered it and posted the source code on the net. It isn't patented. Therefore, RSA is trying to move all its customers to RC5, which is both patented and copyrighted.
Key point: In order to promote RC5, RSA conducts contests that pay people if they can crack it. The first contest used a 56-bit key, took 212 days to crack by http://www.distributed.net/ using a total of roughly 1-million computers trying all possible 35,000,000,000,000,000 combinations. The message was "It is time to move to a longer key length.", and it was encrypted using the key 0x532B744CC20999.
Key point: This allows a spammer with a dial-up account to send e-mail as fast as a high-speed Internet connection, since it is the victim who breaks apart the recpient list and sends each person a separate copy. Therefore, one e-mail goes into the server, thousands come out.
Key point: Relaying can be turned off in the e-mail server configuration. Such configuration will force the server to accept either incoming mail, or outgoing mail, but not incoming e-mail destined back out to the Internet. There are several sites on the Internet that will scan your corporate e-mail server to see if will relay spam.
Resource: Paul Vixie's MAPS http://maps.vix.com/ (MAPS is SPAM spelled backwards).
Analogy: In the 1992 movie Sneakers, the victim uses a voice identification system. Therefore, the heros record the voice of one of the victim's employees, edit it with a computer, then play it back into the voice recognition system.
Key point: It seems the first generation of any security architecture is vulnerable to replay attacks. For example, IPsec was original vulnerable to some replay attacks, even though it had provisions against the most obvious ones.
Key poitn: The anti-replay remedy is to include a timestamp with a message. This then implies that everyone needs to have their clocks synchronized in order to communicate correctly.
Key point: The resource kits contain numerous tools to help system administration. Therefor, these tools are extremely useful for hacking. Any hacker interested in compromising Windows has a copy of the Resource Kit.
Key point: These tools aren't "dangerous". They don't provide any capabilities that hackers couldn't program for themselves. These tools are useless against secured systems.
Key point: The term can be used as a verb. To "root" machine is to break in and obtain root privileges, and their own the machine.
Key point: A rootkit contains many trojaned programs. These programs are used to allow the hacker entry back into the system and to hide the presence of the hacker. For example, a trojaned "ps" command might hide the hacker's sniffer daemon from appearing in the process list. Alternatively, the hacker might trojan an existing daemon like inetd to run a background sniffer.
Key point: The most important trojaned programs are those that deal with gaining access back into the system with a special password. Therefore, trojaned versions of login daemon, su, or telnetd are needed.
Key point: Rootkits often contain setuid programs that normal users can run in order to elevate their privileges to root. Look for these in order to see if your system has been hacked.
Culture: Also called "daemon kits".
Key point: RSA forms the basis for X.509 certificates in web servers and browsers.
Key point: RSA Security charges a hefty license to use the RSA algorithm. However, the patent expires in September of the year 2000. At that time, the number of products using the RSA algorithm are likely to explode.
Key point: An alternative to RSA is the "Diffie-Hellman" algorithm. This is used in many cases, but it is hampered by the fact that many products that could use it (like Netscape and Microsoft browsers) do not; for interoperability you often need to use RSA over DH.
See also: defaults
Key point: Elite hackers write scripts, script-kiddies run scripts.
Misconception: A lot of "scripts" are written in scripting languages like PERL, but a lot are distributed in C/C++ source form as well.
Key point: In theory, setuid programs can only be installed by root, and they are considered as part of the operating system, because they inherently bypass security checks and must verify security themselves. A typical example is the passwd command, which a user runs in order to change his/her password. It must be setuid, because it changes files only root has access to, but yet it must be runnable by users.
Key point: In practice, setuid programs often have bugs that can be exploited by logged in users.
Key point: As part of hardening a system, the administrator should scour the system and remove all unnecessary setuid programs. TODO: show the command to do so.
Key point: Some programs are really setguid which only changes the group context rather than the user context.
Key point: Windows doesn't have the concept of setuid. Instead, RPC is used whereby client programs (run by users) contact server programs to carry out the desired task. For example, in order to change the password, the client program asks the SAM to do it on behalf of the user. Thus, whereas UNIX requires a myriad of client programs to verify credentials and be written securely, Windows only requires a few server programs to do the same.
Key point: In such systems, any computer on the wire can eavesdrop on its neighbors.
Contrast: Most corporations are replacing their shared media nets with switched connections.
Example: DVD movies are encrypted with a randomly generated key. This key is then is then encrypted multiple times with hundreds of different keys. Every DVD player vendor owns one of these keys and imbeds it in their device, thus allows that player to decrypt the movie. (Presumably, if one of the keys is compromised, future movies can be generated without the offending key, causing players based upon that key to become obsolete). However, there is no good way to protect these keys, even though they are in hardware. In late 1999, students in Europe where able to break one of these keys (the Xing software DVD player), and from there they were able to break the majority of the other keys. (These keys only used 40-bit encryption, so breaking one key in the software player allowed a known-plaintext attack).
Key point: This is similar to the "Command Prompt" or incorrectly named "DOS Prompt" on Windows systems.
Key point: Many systems pass filenames along with commands directly to the shell. Hackers can exploit this by sending special shell characters (like the pipe | character) as part of filenames in order to execute their own commands. This is an example of an input validation exploit. Examples of this are web-servers, PERL scripts, and CGI scripts.
Key point: The most popular shell among hackers is probably "bash", the shell from GNU that ships with Linux. (Culture: The original shell on UNIX is known as the "Bourne Shell", named for its creator. The acronym "bash" means "Bourne Again SHell", reflecting that fact that it is a rewrite of that shell).
Key point: Retrieving someones .bash_history file is a common attack against UNIX machines. Several embedded systems have shipped such that the file http://raq.robertgraham.com/~root/.bash_history could be retrieved via the web.
Key point: Marketing forces often mean that companies have to fill their products with useless signatures. Don't be impressed because one product has more signatures than another.
Key point: One of the key goals of hacking is to evade signature detection. Virus writers attempt to encypt their viruses, whereas remote hackers attempt to alter the networking protocol so that it has the same effect, but a different pattern on the wire.
Example: Microsoft's Authenticode allows application developers to sign their programs. Any alteration to the software will result in an invalid signature. Therefore, hackers can't add trojans/viruses to commercial software without it being detected.
Key point: Digital signatures only work if people check them. People rarely check signatures in e-mail or software.
History: SMB was originally developed for DOS machines. It was later upgraded so that OS/2 machines could act as servers for DOS machines. The protocol was later upgraded for Windows (Wfw = Windows for Workgroups) and Windows NT. Still later upgrades have been added for Windows 2000. This constant evolution and need for backwards compatibility has led to numerous security holes within the protocol. The most severe is the need for "LAN Manager" authentication.
Key point: SMB is an appliation layer protocol and can run over many different transports, including TCP/IP. A common problem is that home-users enable SMB over TCP/IP, allowing anybody on the Internet to access their hard-disk. They should instead install a local-only transport such as NetBEUI for SMB, which will allow file access among local machines, but not remote machines across the Internet.
Key point: Virtually all e-mail exchanged on the Internet is through SMTP.
Key point: The most common exploits for SMTP involve spammers trying to relay mail through high-speed mail servers.
Key point: You have be between the sender and the receiver in order to sniff traffic. This is easy in corporations using shared media, but practically impossible with an ISP unless you break into their building or be an employee.
Key point: Sniffers are frequently used as part of automated programs to sift information off the wire, such as clear-text passwords, and sometimes password hashes (to be crack).
Further reading: http://www.robertgraham.com/pubs/sniffing-faq.html.
Key point: The classic example is to pretend to be from a company's computer department and call up a user asking for their password. Sophisticated hacks will first try to make the victim uncomfortable (i.e. "We've detected improper use of your account..."), then offer them the opportunity to be very helpful ("I'm sure we can check this out now and not involve your boss."). The technique often works well in reverse: call up the computer support department and tell them you've lost your password. This works especially well in companies that have policies requiring you to change your password -- people forgetting passwords on really old accounts are frequent, so support departments are deluged with such requests, so it's easy to slip one past them.
Key point: Know as much about your victim as possible. If you are emulating something, try to find the answers to typical questions you will be asked.
Key point: If all else fails, try stupidity. If you are a foreigner, pretend not to speak the language well. Likewise, females have certain advantages in male-dominated cultures.
Key point: Newbies are favorite victims of social-engineering attacks in chat rooms. Hackers go after people who appear to be unsure of themselves online.
Key point: Many hackers do not consider social-engineering a "real" attack because it doesn't require extensive technical knowledge in order to pull off.
Contrast: Other interfaces programmers could use are higher-level abstractions like RPC, or lower-level "raw" interfaces like libnet.
Contrast: Sockets originally came from UNIX, but has been ported to other platforms. In particular, the "WinSock" variant for Windows includes both the UNIX-style functions as well as the Windows-style functions. It is possible to write sockets-based programs that compile for both platforms.
Key point: The name "sockets" comes from the TCP/IP term "socket". A socket is minimum information necessary needed to communication on the network: the source/destination IP address, the source/destination port, and the transport protocol (UDP or TCP).
Key point: SOCKS servers are frequently misconfigured allowing both outside and inside people to use them. This means that if a hacker wants to hide where they come from, the hacker scans the Internet for SOCKS proxies, then funnel their data through the proxies they find. When victims trace back to the hacker's IP address, they find the open SOCKS server instead.
Key point: Abuse through SOCKS servers has become so common on IRC networks that many of them (dalnet, undernet) have begun scanning clients to see if they are running an open SOCKS proxy. They deny access to anybody coming into the networks through such a proxy. Note that users can still use closed proxies (i.e. those available only to internal users).
Key point: SOCKS servers listen by default on TCP port 1080.
Real world: Most browsers support SOCKS, which you can see in the "proxy" settings configuration tab. You can download generic SOCKS clients and servers from http://www.socks.nec.com/.
Analogy: When you send a letter via normal post (snail mail), you write the recipient's name and address on the envelope. You typically also write the sender's name and address as well, so that if there is an error forwarding you letter (e.g. a stamp falls off), they know who sent the letter and can return it. However, you can easily spoof it. For example, someone I know absolutely had to send a letter, but had no stamps. So he simply put the actual recipient's name as the return address section of the envelope and dropped it into the mail box. The letter was returned to sender, which of course arrived at the intended recipient.
Misconception: Most people are interested in spoofing because they think it will allow them to hack a machine in a completely anonymous manner. It doesn't work this way. For example, Mitnick used IP spoofing in order to attack Shimomura's computers, but was caught anyway because spoofing does not truely hide the attacker. The problem is that all responses go back to the sender, so if you've spoofed the sender, you'll never see the responses. Therefore, the spoofing is useless for any normal activity. On the other hand, spoofing can still be useful in situations where seeing the response is not necessary. In the Mitnick instance, two machines trusted each other. Therefore, Mitnick was able to emulate and entire connection between the two machines by "predicting" what all the responses would be. He used this connection to open up something on the victim machine that he could then connect to normally. It was precurser scanning the and the post-spoof connection that Shimomura used to catch Mitnick.
Example: A particularly nasty form of a spoofing is TCP sequence number prediction. Theoretically, you cannot spoof any protocol based upon TCP connections. This is because both sides of a TCP connection choose their own Initial Sequence Number (ISN). In theory, this is a completely random number that cannot be guessed. In practice, it can sometimes be easily guessed. Mitnick used this technique when hacking Shimomura. As of the end of 1999, operating systems such as Linux, WinNT, and Win2k have implemented truely randmon ISNs in order to defeat this type of attack.
Example: In terms of volume of traffic, the most common use of spoofing today is smurf and fraggle attacks. These attacks spoofed packets against amplifiers in order to overload the victim's connection. This is done by sending a single packet to a broadcast address with the victim as the source address. All the machines within the broadcast domain then respond back to the victim, overloading the victim's Internet connection. Since smurfing accounts for more than half the traffic on some backbones, ISPs are starting to take spoofing seriously and have started implementing measures within their routers that verify valid source addresses before passing the packets. As a consequence, spoofing will become incresingly more difficult as time goes on.
Key point: Most of the discussion of spoofing centers around clients masquarading as somebody else. On the other hand, the reverse problem is equally worrisome: hackers can often spoof servers. For example, I post on my website that there is a serious security fix needed to protect yourself while on the web, and point you to http://www.micrsoft.com and hope that you never notice that the URL is misspelled. You would then go to that site (which would be really my server) and download the patch, which would really be a Trojan Horse that I designed in order to break into your computer. This is why server-side certificates are important: some indepdent party validates that the server isn't bogus.
Key point: As the analogy with postal mail shows, many things can be forged, not just the sender's IP address. Most spammers forge their sender's e-mail address in order to avoid all the hate mail they will receive in response. Forging your own sender e-mail address is as simple as reconfiguring your e-mail client -- anybody can do it. (However, there are more secrets to this, which mean you can still be caught by any determined person).
Key point: Web servers have a certificate signed by a trusted certificate authority (CA). This certificate allows the client and the server to generate random keys for the session and to exchange them securely (to defend against man-in-the-middle attacks). The generated random key is used to encrypt the rest of the contents of the connection, usually using RC4. U.S. export controls attempts to limit products used abroad to only 40-bits of key length, which can easily be broken.
Key point: In SSL, the server first authenticates itself with the client (a process that makes it more likely that e-commerce vendors are reputable). Therefore, if you want to set up your own SSL-based web server, you need to get a signed certificate from a CA. Furthermore, if you are outside the U.S., you will find it difficult to find one for 128-bits, though the Chaos Computer Club in Germany manages nicely.
Key point: The chief reason SSL isn't used more widely is because it creates a huge performance hit on servers due to the need to encrypt/decrypt everything. This is changing as new devices are becoming available that offloads this process. CrypoSwitch
History: SSL was originally developed by Netscape to promote e-commerce. It is also known under the IETF standard name of TLS (Transport Layer Security).
Contrast: In the 1980s, SunOS was based upon BSD. In the 1990s, Sun replaced SunOS with Solaris, a System V based operating system.
Key point: The word "SunOS" refers to SunOS version 4. The word "Solaris" refers to SunOS version 5.
Key point: The last major version of SunOS was 4.1.3, and continues to be popular (in much the same way that DOS and Win 3.1 continues to be installed on new machines). As a result, there are thousands of SunOS machines still out there that haven't been patched and which are susceptable to old exploits.
This process is generally called "swapping" or "paging". The word "swap" reflects the fact that inactive blocks of memory are being switched with active blocks from the disk. The word "paging" reflects the fact that a common name for a block of memory is "page". The name of the file on the disk that an OS uses for swapping is called the "swapfile" or "pagefile".
Key point: A lot of security depends upon the fact that memory is secure: the OS protects applications from reading other application's memory, and that when the computer is turned off, the memory is erased. Therefore, applications can safely store passwords in clear-text in memory. Swapping defeats this, because the memory pages that store the passwords may have been swapped to the disk. Someone with physical access to the machine can turn it off, steal the disk, and run the pagefile through analysis programs in order possibly retrieve passwords.
Key point: TCP is "connection oriented". This means the three-way handshake must be completed before any data can be sent across the connection. This makes IP address spoofing impossible without sequence number prediction.
Key point: TCP creates a virtual "byte stream" for applications. Therefore, applications that send/receive data must create their own boundaries, such as length encoding the data, or send text data a line at a time. However, in practice, applications do indeed send data aligned on packet boundaries. Most network-based intrusion detection systems depend upon these boundaries in order to work correctly. Therefore, they can easily be evaded by custom written scripts that misalign the data. The applications don't see any difference, but the NIDS see something completely different go across the wire that no longer matches their signatures.
Contrast: There are two transport protocols: TCP and UDP. Whereas TCP is connection-oriented, UDP is connectionless, meaning UDP-based applications are easily spoofed.
Key point: Telnet can open up a raw TCP connection to any port in order to allow a hacker to interact directly with text-based protocols.
Example: Telnet to your local SMTP using a command that looks like telnet smtp.example.com 25. The first parameter should be your own mail server, whereas the second parameter indicates which port to connect to (other than the default port 23). Now type in the text as you see it below:
HELO foo.example.com MAIL FROM: firstname.lastname@example.org RCPT TO: email@example.com DATA this data will appear in the contents of the e-mail message .This will send the indicated e-mail message with the From: and To: addresses with the indicated content.
Key point: When abusing Telnet in this fashion, you cannot see the echoed characters, nor can you edit what you type by using the backspace key. Remember that the service on the other end thinks you are a program, so you shouldn't need to see the characters you type, and you should type these characters correctly the first time.
Key point: Some intrusion detection systems, like Network ICE, character-by-character activity instead of the expected line-by-line activity.
Contrast: The netcat tool provides the same ability to open a raw TCP connection, but sends a line at a time, echoes the characters back so you can see what you type, and allows you to edit the line before sending it. It also allows you to recieve incoming connections, which might be useful when hacking FTP.
Key point: The word "van Eck monitoring" refers to the ability to remotely view a terminal/CRT from its radiation (see Phrack 44-11). Ross Anderson and Markus Kuhn have come up with an innovative technique of producing fonts that remove the high-frequency information, and thus severely reduce the ability to remotely view text on the screen.
Resources: See The Complete, Unofficial TEMPEST Information Page at http://www.eskimo.com/~joelm/tempest.html.
Misconception: Many people desfcribe TFTP as simply a trivial version of FTP. This misses the point. The purpose of TFTP is not to reduce the complexity of file transfer, but to reduce the complexity of the the underlying TCP/IP stack so that it can fit inside boot ROMs.
Key point: TFTP is almost always used with BOOTP. BOOTP first configures the device, then TFTP transfers the boot image.
Key point: Numerous systems come with unnecessary TFTP servers. Many TFTP servers have bugs, like the directory traversal problem or buffer overflows. As a consequence, many systems can be exploited with TFTP even though virtually nobody really uses it.
Key point: The word can be used as a verb. To trojan a program is to add subversive functionality to an existing program. For example, a trojaned login program might be programmed to accept a certain password for any user's account that the hacker can use to log back into the system at any time. Rootkits often contain a suite of such trojaned programs.
Key point: Users can often break into a system by leaving behind trojaned command programs in directories (like their own directory or the /tmp directory). If you copy your own ls program to the /tmp directory, and somebody else does a cd /tmp then an ls, that user will run your program with their own privileges. This is especially dangerous against root, which is why the local directory should not be part of the search path for the root account.
Example: In the movies you often see that nuclear weapons have two seperate keys in order to unlock them. The locks are placed in positions further appart than a single person can reach. The keys must be turned at the same time in order to unlock the system.
Example: Really important passwords (such as those protecting private keys) are often given in pieces -- different pieces to different people. This requires that multiple people to be present in order to log on.
Key point: There really is no "UNIX", but just various implementations designed along the same guidelines. Different versions of UNIX are more or less related, and there is extensive cross-germination of ideas, so that something good that appears in one will eventually migrate to others.
Contrast: There have been two main branches of UNIX: SVR4 (System V Release 4) and BSD (Berkeley Standard Distribution). Many security issues depend upon which base the system was derived.
Example: Sun Solaris, IBM AIX, SCO, SGI Irix, Apple A/UX, BSD, HP/UX.
Key point: UNIX is case-sensitive, whereas Windows and Macintosh are "case-insensitive" but "case-preserving". Windows has a compatibility mode that allows case-sensitivity, which can sometimes be exploited with other techniques in order to compromise the system.
Key point: The BSD branch has spawned many open-source variants, such as FreeBSD and OpenBSD. OpenBSD is considered one of the more secure versions of UNIX. Security experts spend the most time on OpenBSD in order to clean up bugs like buffer-overflows. However, in 1999, the dramatic rise of hacking and publication of bugs has led to a hightened awareness of these problems, which may lead to other systems becoming equally scoured for bugs.
How to: In order to harden UNIX, you generally do the following:
Key point: This encoding mechanism can be used to alter the signature of a hacker attack via web-based protocols. Such encoding can be used to evade detection by lightweight intrusion detection systems that are unable to "normalize" the URL.
Example: The Microsoft webserver in their ASP server-side scripts such that a hacker could append a dot to the end of the URL in order to read the script contents rather than executing the script. Microsoft created a patch, but hackers soon found they could evade the patch by URL-encoding the dot (appending a %2E to the end of the scrip rather than a dot). Examples:
|http://www.robertgraham.com/sample.asp.||Attempt to read script rather than executing it.|
|http://www.robertgraham.com/sample.asp%2E||URL-encoding in order to evade patch.|
|http://www.robertgraham.com/sample.%61sp%2E||Fruther URL-encoding in order to evade intrusion detection systems.|
Key point: The USENET Death Penalty is often applied to NNTP servers in order to stop the flood of spam. It is often applied to ISPs who allow users to send lots of spam or allow their servers to be hijacked.
Key point: Even though it is rarely used today, uucp accounts and services are often enabled on UNIX machine in such a way that they can be exploited in order to break into the machine.
Analogy: A biological virus is not a "living" thing. Instead, it is simply a strand of DNA. When it enters a living cell, it takes control of the cell forcing it to generate duplicate copies of the original DNA strand. In much the same way, a computer virus hijacks the computer forcing it to generate duplicate copies of the original virus. Computer viruses are so common because humans do not practice sufficient cyber-hygene when exchanging files.
Key point: An "anti-virus" programs scans the disks on your system hunting down those files that have signatures indicative of infected files. Since file-scanning technology is generic, most anit-virus programs also scan for other hostile content, such as trojans.
Contrast: A virus is a program that spreads itself, whereas a trojan spread by a hacker. In other words, you catch a virus because you ran some program a friend gave you. You catch a trojan because you ran some program the hacker gave you. A common phrase people make is "The hacker put a virus on my disk". What they really mean is that the hacker put some sort of hostile program (usually a remote access Trojan) which is not any sort of virus at all. While the hacker's program will indeed try to ensconce itself deeply in the system, it will not try to spread to other systems.
Key point: VPNs are used for employee-company and company-company connections.
Key point: One way that an employee can connect to a company is to put a modem in the machine and dial directly to modems inside the corporation. This is expensive due to long distance charges. But think for a moment that the employee can purchase two modems to put in the machine, and while dialed up to the corporation, the employee also dials up the Internet. This would mean that the employee has two active network connections: one to the corporation, one to the Internet. A VPN is the same thing, only the corporate connection and modem are "virtual".
Key point: Vendors claim that when the VPN is active, that the previous Internet access is disabled and all further communication goes through the corpration. Therefore, if the user wants to browse the web while the VPN is active, the user must browse through firewalls/proxies inside the corporation then back out to the web. However, this is just a bit of sleight-of-hand: while it appears to the user that normal Internet communication has been disabled, in reality it has only been "hidden": a hacker can still compromise the machine from the Internet.
Key point: VPN puts the connection on the company's internal network, inside the firewall. Therefore, if a hacker compromises someone's machine who uses VPN, then the hacker has easy access to the inside of a hardened corporate environment.
Tool: The program ToneLoc for DOS is one of the most popular among hackers for this purpose.
Key point: Many corporate desktops run PCAnywhere. This allows employees to access their desktop computers from home without the firewall-nazis blocking access. They also install PCAnywhere without those pesky passwords. Consequence: hackers who wardial often come up with PCAnywhere machines that they can easily connect to and break into companies.
Key point: Other popular applications that pick up dialup lines are Windows RAS servers, Laplink, and telnet-like terminal servers.
Key point: On Windows, trailing dots on filenames are ignored. This means the filenames "foo" and "foo." are the same. However, most applications treat these two filenames as representing different files. Hackers can sometimes exploit this difference. For example, on older versions of ISS, this could often be used to read the contents of scripts rather than running them. Being able to read the script isn't necessarily a security breach, but a hacker could use the script in order find other ways of breaking into the system.
Key point: Microsoft has obtained C2 certification for Windows NT. This doesn't mean that WinNT is more secure than other operating systems, but it does mean that WinNT has features required to harden a system according to this government specification.
Example: In the late 1980s, the Morris Worm essentially shutdown the Internet for a day. TODO
Key point: X Windows goes in the "wrong" direction. When you log into an X Windows host, the host opens a connection back to the display. As a consequence, it is very useful as a back-channel. In particular, the program xterm provides a raw command-prompt from which a hacker can interact with just as if they had telnetted to the machine.
plaintext 11100101 01110101 key 00001111 00001111 ------------------ ciphertext 11101010 01111010Moreover, XOR has the interesting property that XORing by the same pattern twice results in the original pattern:
ciphertext 11101010 01111010 key 00001111 00001111 ------------------ plaintext 11100101 01110101Therefore, you can think of XOR as an extremely weak encryption algorithm. The above example shows using XOR as a way of encrypting the original data with the 8-bit key of "00001111". Many products use this technique to obfuscate data. However, it is extremely easy to recover the original key via a known plaintext attack, as show below:
ciphertext 11101010 01111010 plaintext 11100101 01110101 ------------------ key 00001111 00001111
Key point: XOR is a common mathematical operation used in cryptographic algorithms. In fact, the only 100% secure form of encryption is XORing against a one-time pad. Also, any hash algorithm can be converted into an encryption algorithm though a clever use of XOR.