#127 (935 lines in body): Delivery-Date: 26 September 1986 23:06 edt Delivery-By: Network_Server.Daemon (NEUMANN@CSL.SRI.COM@ATHENA) Date: Friday, 26 September 1986 16:33 edt From: Peter G. Neumann <Neumann at CSL.SRI.COM> Subject: SUMMARY OF UNIX BREAKIN MESSAGES To: Saltzer at ATHENA.MIT.EDU In-Reply-To: Message from "Jerome H. Saltzer <Saltzer@ATHENA.MIT.EDU>" of Fri 26 Sep 86 13:29:11-PDT RISKS@CSL.SRI.COM RISKS-FORUM Digest, Summary of messages on UNIX breakins. THIS FILE IS AVAILABLE FOR FTP FROM CSL.SRI.COM <RISKS>RISKS.REID. RISKS-LIST: RISKS-FORUM Digest, Tuesday, 16 September 1986 Volume 3 : Issue 56 From: reid@decwrl.DEC.COM (Brian Reid) Date: 16 Sep 1986 1519-PDT (Tuesday) To: Peter G. Neumann <Neumann@csl.sri.com> [FOR RISKS] Subject: Massive UNIX breakins at Stanford Lessons learned from a recent rash of Unix computer breakins Introduction A number of Unix computers in the San Francisco area have recently been plagued with breakins by reasonably talented intruders. An analysis of the breakins (verified by a telephone conversation with the intruders!) show that the networking philosophy offered by Berkeley Unix, combined with the human nature of systems programmers, creates an environment in which breakins are more likely, and in which the consequences of breakins are more dire than they need to be. People who study the physical security of buildings and military bases believe that human frailty is much more likely than technology to be at fault when physical breakins occur. It is often easier to make friends with the guard, or to notice that he likes to watch the Benny Hill show on TV and then wait for that show to come on, than to try to climb fences or outwit burglar alarms. Summary of Berkeley Unix networking mechanism: The user-level networking features are built around the principles of "remote execution" and "trusted host". For example, if you want to copy a file from computer A to computer B, you type the command rcp A:file B:file If you want to copy the file /tmp/xyz from the computer that you are now using over to computer C where it will be called /usr/spool/breakin, you type the command rcp /tmp/xyz C:/usr/spool/breakin The decision of whether or not to permit these copy commands is based on "permission" files that are stored on computers A, B, and C. The first command to copy from A to B will only work if you have an account on both of those computers, and the permission file stored in your directory on both of those computers authorizes this kind of remote access. Each "permission file" contains a list of computer names and user login names. If the line "score.stanford.edu reid" is in the permission file on computer "B", it means that user "reid" on computer "score.stanford.edu" is permitted to perform remote operations such as rcp, in or out, with the same access privileges that user "reid" has on computer B. How the breakins happened. One of the Stanford campus computers, used primarily as a mail gateway between Unix and IBM computers on campus, had a guest account with user id "guest" and password "guest". The intruder somehow got his hands on this account and guessed the password. There are a number of well-known security holes in early releases of Berkeley Unix, many of which are fixed in later releases. Because this computer is used as a mail gateway, there was no particular incentive to keep it constantly up to date with the latest and greatest system release, so it was running an older version of the system. The intruder instantly cracked "root" on that computer, using the age-old trojan horse trick. (He had noticed that the guest account happened to have write permission into a certain scratch directory, and he had noticed that under certain circumstances, privileged jobs could be tricked into executing versions of programs out of that scratch directory instead of out of the normal system directories). Once the intruder cracked "root" on this computer, he was able to assume the login identity of everybody who had an account on that computer. In particular, he was able to pretend to be user "x" or user "y", and in that guise ask for a remote login on other computers. Sooner or later he found a [user,remote-computer] pair for which there was a permission file on the other end granting access, and now he was logged on to another computer. Using the same kind of trojan horse tricks, he was able to break into root on the new computer, and repeat the process iteratively. In most cases the intruder left trojan-horse traps behind on every computer that he broke into, and in most cases he created login accounts for himself on the computers that he broke into. Because no records were kept, it is difficult to tell exactly how many machines were penetrated, but the number could be as high as 30 to 60 on the Stanford campus alone. An intruder using a similar modus operandi has been reported at other installations. How "human nature" contributed to the problem The three technological entry points that made this intrusion possible were: * The large number of permission files, with entirely too many permissions stored in them, found all over the campus computers (and, for that matter, all over the ARPAnet). * The presence of system directories in which users have write permission. * Very sloppy and undisciplined use of search paths in privileged programs and superuser shell scripts. Permissions: Berkeley networking mechanism encourages carelessness. The Berkeley networking mechanism is very very convenient. I use it all the time. You want to move a file from one place to another? just type "rcp" and it's there. Very fast and very efficient, and quite transparent. But sometimes I need to move a file to a machine that I don't normally use. I'll log on to that machine, quickly create a temporary permission file that lets me copy a file to that machine, then break back to my source machine and type the copy command. However, until I'm quite certain that I am done moving files, I don't want to delete my permission file from the remote end or edit that entry out of it. Most of us use display editors, and oftentimes these file copies are made to remote machines on which the display editors don't always work quite the way we want them to, so there is a large nuisance factor in running the text editor on the remote end. Therefore the effort in removing one entry from a permission file--by running the text editor and editing it out--is high enough that people don't do it as often as they should. And they don't want to *delete* the permission file, because it contains other entries that are still valid. So, more often than not, the permission files on rarely-used remote computers end up with extraneous permissions in them that were installed for a one-time-only operation. Since the Berkeley networking commands have no means of prompting for a password or asking for the name of a temporary permission file, everybody just edits things into the permanent permission file. And then, of course, they forget to take it out when they are done. Write permission in system directories permits trojan horse attacks. All software development is always behind schedule, and programmers are forever looking for ways to do things faster. One convenient trick for reducing the pain of releasing new versions of some program is to have a directory such as /usr/local/bin or /usr/stanford/bin or /usr/new in which new or locally-written versions of programs are kept, and asking users to put that directory on their search paths. The systems programmers then give themselves write access to that directory, so that they can intall a new version just by typing "make install" rather than taking some longer path involving root permissions. Furthermore, it somehow seems more secure to be able to install new software without typing the root password. Therefore it is a nearly-universal practice on computers used by programmers to have program directories in which the development programmers have write permission. However, if a user has write permission in a system directory, and if an intruder breaks into that user's account, then the intruder can trivially break into root by using that write permission to install a trojan horse. Search paths: people usually let convenience dominate caution. Search paths are almost universally misused. For example, many people write shell scripts that do not specify an explicit search path, which makes them vulnerable to inheriting the wrong path. Many people modify the root search path so that it will be convenient for systems programmers to use interactively as the superuser, forgetting that the same search path will be used by system maintenance scripts run automatically during the night. It is so difficult to debug failures that are caused by incorrect search paths in automatically-run scripts that a common "repair" technique is to put every conceivable directory into the search path of automatically-run scripts. Essentially every Unix computer I have ever explored has grievous security leaks caused by underspecified or overlong search paths for privileged users. Summary conclusion: Wizards cause leaks The people who are most likely to be the cause of leaks are the wizards. When something goes wrong on a remote machine, often a call goes in to a wizard for help. The wizard is usually busy or in a hurry, and he often is sloppier than he should be with operations on the remote machine. The people who are most likely to have permission files left behind on stray remote machines are the wizards who once offered help on that machine. But, alas, these same wizards are the people who are most likely to have write access to system directories on their home machines, because it seems to be in the nature of wizards to want to collect as many permissions as possible for their accounts. Maybe that's how they establish what level of wizard that they are. The net result is that there is an abnormally high probability that when an errant permission file is abused by an intruder, that it will lead to the account of somebody who has an unusually large collection of permissions on his own machine, thereby making it easier to break into root on that machine. Conclusions. My conclusions from all this are these: * Nobody, no matter how important, should have write permission into any directory on the system search path. Ever. * Somebody should carefully re-think the user interface of the Berkeley networking mechanisms, to find ways to permit people to type passwords as they are needed, rather than requiring them to edit new permissions into their permissions files. * The "permission file" security access mechanism seems fundamentally vulnerable. It would be quite reasonable for a system manager to forbid the use of them, or to drastically limit the use of them. Mechanized checking is easy. * Programmer convenience is the antithesis of security, because it is going to become intruder convenience if the programmer's account is ever compromised. This is especially true in setting up the search path for the superuser. Lament I mentioned in the introduction that we had talked to the intruders on the telephone. To me the most maddening thing about this intrusion was not that it happened, but that we were unable to convince any authorities that it was a serious problem, and could not get the telephone calls traced. At one point an intruder spent 2 hours talking on the telephone with a Stanford system manager, bragging about how he had done it, but there was no way that the call could be traced to locate him. A few days later, I sat there and watched the intruder log on to one Stanford comptuer, and I watched every keystroke that he typed on his keyboard, and I watched him break in to new directories, but there was nothing that I could do to catch him because he was coming in over the telephone. Naturally as soon as he started to do anything untoward I blasted the account that he was using and logged him off, but sooner or later new intruders will come along, knowing that they will not be caught because what they are doing is not considered serious. It isn't necessarily serious, but it could be. I don't want to throw such people in jail, and I don't want to let them get away either. I just want to catch them and shout at them and tell them that they are being antisocial. Brian Reid DEC Western Research and Stanford University ------------------------------ RISKS-LIST: RISKS-FORUM Digest, Wednesday, 17 Sept 1986 Volume 3 : Issue 58 From: davy@ee.ecn.purdue.edu (Dave Curry) To: risks@csl.sri.com Cc: reid@decwrl.dec.com Subject: Massive UNIX breakins Date: Wed, 17 Sep 86 08:01:03 EST Brian - I feel for you, I really do. Breakins can be a real pain in the neck, aside from being potentially hazardous to your systems. And, we too have had trouble convincing the authorities that anything serious is going on. (To their credit, they have learned a lot and are much more responsive now than they were a few years ago.) I do have a couple of comments though. Griping about the Berkeley networking utilities is well and good, and yes, they do have their problems. However, I think it really had little to do with the initial breakins on your system. It merely compounded an already exisiting breakin several fold. Two specific parts of your letter I take exception to: One of the Stanford campus computers, used primarily as a mail gateway between Unix and IBM computers on campus, had a guest account with user id "guest" and password "guest". The intruder somehow got his hands on this account and guessed the password. Um, to put it mildly, you were asking for it. "guest" is probably the second or third login name I'd guess if I were trying to break in. It ranks right up there with "user", "sys", "admin", and so on. And making the password to "guest" be "guest" is like leaving the front door wide open. Berkeley networking had nothing to do with your initial breakin, leaving an obvious account with an even more obvious password on your system was the cause of that. There are a number of well-known security holes in early releases of Berkeley Unix, many of which are fixed in later releases. Because this computer is used as a mail gateway, there was no particular incentive to keep it constantly up to date with the latest and greatest system release, so it was running an older version of the system. Once again, you asked for it. If you don't plug the holes, someone will come along and use them. Again Berkeley networking had nothing to do with your intruder getting root on your system, that was due purely to neglect. Granted, once you're a super-user, the Berkeley networking scheme enables you to invade many, many accounts on many, many machines. Don't get me wrong. I'm not trying to criticize for the sake of being nasty here, but rather I'm emphasizing the need for enforcing other good security measures: 1. Unless there's a particularly good reason to have one, take all "generic" guest accounts off your system. Why let someone log in without identifying himself? 2. NEVER put an obvious password on a "standard" account. This includes "guest" on the guest account, "system" on the root account, and so on. Enforcing this among the users is harder, but not impossible. We have in the past checked all the accounts on our machines for stupid passwords, and informed everyone whose password we found that they should change it. As a measure of how simple easy passwords make things, we "cracked" about 400 accounts out of 10,000 in one overnight run of the program, trying about 12 passwords per account. Think what we could have done with a sophisticated attack. 3. FIX SECURITY HOLES. Even on "unused" machines. It's amazing how many UNIX sites have holes wide open that were plugged years ago. I even found a site still running with the 4.2 distributed sendmail a few months ago... 4. Educate your police and other authorities about what's going on. Invite them to come learn about the computer. Give them an account and some documentation. The first time we had a breakin over dialup (1982 or so), it took us three days to convince the police department that we needed the calls traced. Now, they understand what's going on, and are much quicker to respond to any security violations we deem important enough to bring to their attention. The Dean of Students office is now much more interested in handling cases of students breaking in to other students' accounts; several years ago their reaction was "so what?". This is due primarily to our people making an effort to educate them, although I'm sure the increased attention computer security has received in the media (the 414's, and so on) has had an effect too. --Dave Curry Purdue University Engineering Computer Network ------------------------------ From: reid@decwrl.DEC.COM (Brian Reid) Date: 17 Sep 1986 0729-PDT (Wednesday) To: davy@ee.ecn.purdue.edu (Dave Curry) Cc: risks@csl.sri.com Subject: Massive UNIX breakins The machine on which the initial breakin occurred was one that I didn't even know existed, and over which no CS department person had any control at all. The issue here is that a small leak on some inconsequential machine in the dark corners of campus was allowed to spread to other machines because of the networking code. Security is quite good on CSD and EE machines, because they are run by folks who understand security. But, as this episode showed, that wasn't quite good enough. ------------------------------ RISKS-LIST: RISKS-FORUM Digest, Saturday, 20 September 1986 Volume 3 : Issue 60 Date: Thu, 18 Sep 86 09:12:59 cdt From: "Scott E. Preece" <preece%ccvaxa@GSWD-VMS.ARPA> To: RISKS@CSL.SRI.COM Subject: Re: Massive UNIX breakins at Stanford > From: reid@decwrl.DEC.COM (Brian Reid) The machine on which the initial > breakin occurred was one that I didn't even know existed, and over > which no CS department person had any control at all. The issue here is > that a small leak on some inconsequential machine in the dark corners > of campus was allowed to spread to other machines because of the > networking code. Security is quite good on CSD and EE machines, because > they are run by folks who understand security. But, as this episode > showed, that wasn't quite good enough. ---------- No you're still blaming the networking code for something it's not supposed to do. The fault lies in allowing an uncontrolled machine to have full access to the network. The NCSC approach to networking has been just that: you can't certify networking code as secure, you can only certify a network of machines AS A SINGLE SYSTEM. That's pretty much the approach of the Berkeley code, with some grafted on protections because there are real-world situations where you have to have some less-controlled machines with restricted access. The addition of NFS makes the single-system model even more necessary. scott preece, gould/csd - urbana, uucp: ihnp4!uiucdcs!ccvaxa!preece ------------------------------ RISKS-LIST: RISKS-FORUM Digest, Monday, 22 September 1986 Volume 3 : Issue 62 Date: Mon, 22 Sep 86 11:04:16 EDT To: RISKS FORUM (Peter G. Neumann -- Coordinator) <RISKS@CSL.SRI.COM> Subject: Massive UNIX breakins at Stanford From: Jerome H. Saltzer <Saltzer@ATHENA.MIT.EDU> In RISKS-3.58, Dave Curry gently chastises Brian Reid: > . . . you asked for it. . . Berkeley networking had nothing to > do with your intruder getting root on your system, that was due purely > to neglect. Granted, once you're a super-user, the Berkeley networking > scheme enables you to invade many, many accounts on many, many machines. And in RISKS-3.59, Scott Preece picks up the same theme, suggesting that Stanford failed by not looking at the problem as one of network security, and, in the light of use of Berkeley software, not enforcing a no-attachment rule for machines that don't batten down the hatches. These two technically- and policy-based responses might be more tenable if the problem had occurred at a military base. But a university is a different environment, and those differences shed some light on environments that will soon begin to emerge in typical commercial and networked home computing settings. And even on military bases. There are two characteristics of the Stanford situation that RISK-observers should keep in mind: 1. Choice of operating system software is made on many factors, not just the quality of the network security features. A university has a lot of reasons for choosing BSD 4.2. Having made that choice, the Berkeley network code, complete with its casual approach to network security, usually follows because the cost of changing it is high and, as Brian noted, its convenience is also high. 2. It is the nature of a university to allow individuals to do their own thing. So insisting that every machine attached to a network must run a certifably secure-from-penetration configuration is counter-strategic. And on a campus where there may be 2000 privately administered Sun III's, MicroVAX-II's, and PC RT's all running BSD 4.2, it is so impractical as to be amusing to hear it proposed. Even the military sites are going to discover soon that configuration control achieved by physical control of every network host is harder than it looks in a world of engineering workstations. Brian's comments are very thoughtful and thought-provoking. He describes expected responses of human beings to typical current-day operating system designs. The observations he makes can't be dismissed so easily. Jerry Saltzer ------------------------------ Date: Mon, 22 Sep 1986 23:03 EDT From: Rob Austein <SRA@XX.LCS.MIT.EDU> To: RISKS@CSL.SRI.COM Subject: Massive UNIX breakins at Stanford I have to take issue with Scott Preece's statement that "the fault lies in allowing an uncontrolled machine to have full access to the network". This may be a valid approach on a small isolated network or in the military, but it fails horribly in the world that the rest of us have to live in. For example, take a person (me) who is (theoreticly) responsible for what passes for security on up to half a dozen mainframes at MIT (exact number varies). Does he have any control over what machines are put onto the network even across the street on the MIT main campus? Hollow laugh. Let alone machines at Berkeley or (to use our favorite local example) the Banana Junior 6000s belonging to high school students in Sunnyvale, California. As computer networks come into wider use in the private sector, this problem will get worse, not better. I'm waiting to see when AT&T starts offering a long haul packet switched network as common carrier. Rule of thumb: The net is intrinsicly insecure. There's just too much cable out there to police it all. How much knowledge does it take to tap into an ethernet? How much money? I'd imagine that anybody with a BS from a good technical school could do it in a week or so for under $5000 if she set her mind to it. As for NFS... you are arguing my case for me. The NFS approach to security seems bankrupt for just this reason. Same conceptual bug, NFS simply agravates it by making heavier use of the trusted net assumption. Elsewhere in this same issue of RISKS there was some discussion about the dangers of transporting passwords over the net (by somebody other than Scott, I forget who). Right. It's a problem, but it needn't be. Passwords can be tranmitted via public key encryption or some other means. The fact that most passwords are currently transmitted in plaintext is an implementation problem, not a fundamental design issue. A final comment and I'll shut up. With all this talk about security it is important to keep in mind the adage "if it ain't broken, don't fix it". Case in point. We've been running ITS (which has to be one of the -least- secure operating systems ever written) for something like two decades now. We have surprisingly few problems with breakins on ITS. Seems that leaving out all the security code made it a very boring proposition to break in, so almost nobody bothers (either that or they are all scared off when they realize that the "command processor" is an assembly language debugger ... can't imagine why). Worth thinking about. The price paid for security may not be obvious. --Rob Austein <SRA@XX.LCS.MIT.EDU> ------------------------------ Date: Mon 22 Sep 86 11:07:04-PDT From: Andy Freeman <ANDY@Sushi.Stanford.EDU> Subject: Massive UNIX breakins at Stanford To: RISKS@CSL.SRI.COM, preece%ccvaxa@GSWD-VMS.ARPA Scott E. Preece <preece%ccvaxa@GSWD-VMS.ARPA> writes in RISKS-3.60: reid@decwrl.DEC.COM (Brian Reid) writes: The issue here is that a small leak on some [unknown] inconsequential machine in the dark corners of campus was allowed to spread to other machines because of the networking code. No, you're still blaming the networking code for something it's not supposed to do. The fault lies in allowing an uncontrolled machine to have full access to the network. The NCSC approach to networking has been just that: you can't certify networking code as secure, you can only certify a network of machines AS A SINGLE SYSTEM. That's pretty much the approach of the Berkeley code, with some grafted on protections because there are real-world situations where you have to have some less-controlled machines with restricted access. The addition of NFS makes the single-system model even more necessary. Then NCSC certification means nothing in many (most?) situations. A lot of networks cross adminstrative boundaries. (The exceptions are small companies and military installations.) Even in those that seemingly don't, phone access is often necessary. Network access should be as secure as phone access. Exceptions may choose to disable this protection but many of us won't. (If Brian didn't know about the insecure machine, it wouldn't have had a valid password to access his machine. He'd also have been able to choose what kind of access it had.) The only additional problem that networks pose is the ability to physically disrupt other's communication. -andy [There is some redundancy in these contributions, but each makes some novel points. It is better for you to read selectively than for me to edit. PGN] ------------------------------ Date: 22 Sep 1986 16:24-CST From: "Scott E. Preece" <preece%mycroft@GSWD-VMS.ARPA> Subject: Massive UNIX breakins at Stanford (RISKS-3.60) To: ANDY@SUSHI.STANFORD.EDU, RISKS%CSL.SRI.COM@CSNET-RELAY.ARPA Andy Freeman writes [in response to my promoting the view of a network as a single system]: > Then NCSC certification means nothing in many (most?) situations. -------- Well, most sites are NOT required to have certified systems (yet?). If they were, they wouldn't be allowed to have non-complying systems. The view as a single system makes the requirements of the security model feasible. You can't have anything in the network that isn't part of your trusted computing base. This seems to be an essential assumption. If you can't trust the code running on another machine on your ethernet, then you can't believe that it is the machine it says it is, which violates the most basic principles of the NCSC model. (IMMEDIATE DISCLAIMER: I am not part of the group working on secure operating systems at Gould; my knowledge of the area is superficial, but I think it's also correct.) [NOTE: The word "NOT" in the first line of this paragraph was interpolated by PGN as the presumed intended meaning.] -------- Network access should be as secure as phone access. Exceptions may choose to disable this protection but many of us won't. (If Brian didn't know about the insecure machine, it wouldn't have had a valid password to access his machine. He'd also have been able to choose what kind of access it had.) The only additional problem that networks pose is the ability to physically disrupt other's communication. -------- Absolutely, network access should be as secure as phone access, IF YOU CHOOSE TO WORK IN THAT MODE. Our links to the outside world are as tightly restricted as our dialins. The Berkeley networking software is set up to support a much more integrated kind of network, where the network is treated as a single system. For our development environment that is much more effective. You should never allow that kind of access to a machine you don't control. Never. My interpretation of the original note was that the author's net contained machines with trusted-host access which should not have had such access; I contend that that represents NOT a failing of the software, but a failing of the administration of the network. scott preece gould/csd - urbana, uucp: ihnp4!uiucdcs!ccvaxa!preece ------------------------------ RISKS-LIST: RISKS-FORUM Digest Wednesday, 24 September 1986 Volume 3 : Issue 63 Date: Tue, 23 Sep 86 09:16:21 cdt From: "Scott E. Preece" <preece%ccvaxa@GSWD-VMS.ARPA> To: RISKS@CSL.SRI.COM Subject: Massive UNIX breakins at Stanford [This was an addendum to Scott's contribution to RISKS-3.61. PGN] I went back and reviewed Brian Reid's initial posting and found myself more in agreement than disagreement. I agree that the Berkeley approach offers the unwary added opportunities to shoot themselves in the foot and that local administrators should be as careful of .rhosts files as they are of files that are setuid root; they should be purged or justified regularly. I also agree that it should be possible for the system administrator to turn off the .rhosts capability entirely, which currently can only be done in the source code and that it would be a good idea to support password checks (as a configuration option) on rcp and all the other remote services. scott preece, gould/csd - urbana, uucp: ihnp4!uiucdcs!ccvaxa!preece ------------------------------ Date: Tue, 23 Sep 86 08:41:29 cdt From: "Scott E. Preece" <preece%ccvaxa@GSWD-VMS.ARPA> To: RISKS@CSL.SRI.COM Subject: Re: Massive UNIX breakins at Stanford > From: Rob Austein <SRA@XX.LCS.MIT.EDU> > I have to take issue with Scott Preece's statement that "the fault lies > in allowing an uncontrolled machine to have full access to the network"... I stand by what I said, with the important proviso that you notice the word "full" in the quote. I took the description in the initial note to mean that the network granted trusted access to all machines on the net. The Berkeley networking code allows the system administrator for each machine to specify what other hosts on the network are to be treated as trusted and which are not. The original posting spoke of people on another machine masquerading as different users on other machines; that is only possible if the (untrustworthy) machine is in your hosts.equiv file, so that UIDs are equivalenced for connections from that machine. If you allow trusted access to a machine you don't control, you get what you deserve. Also note that by "the network" I was speaking only of machines intimately connected by ethernet or other networking using the Berkeley networking code, not UUCP or telephone connections to which normal login and password checks apply. The description in the original note STILL sounds to me like failure of administration rather than failure of the networking code. scott preece [OK. Enough on that. The deeper issue is that most operating systems are so deeply flawed that you are ALWAYS at risk. Some tentative reports of Trojan horses discovered in RACF/ACF2 systems in Europe are awaiting details and submission to RISKS. But their existence should come as no surprise. Any use of such a system in a hostile environment could be considered a failure of administration. But it is also a shortcoming of the system itself... PGN] ------------------------------ RISKS-LIST: RISKS-FORUM Digest Wednesday, 25 September 1986 Volume 3 : Issue 65 Date: Mon 22 Sep 86 17:09:27-PDT From: Andy Freeman <ANDY@SUSHI.STANFORD.EDU> Subject: UNIX and network security again To: preece%mycroft@GSWD-VMS.ARPA cc: RISKS%CSL.SRI.COM@CSNET-RELAY.ARPA preece%mycroft@gswd-vms.ARPA (Scott E. Preece) writes: If you can't trust the code running on another machine on your ethernet, then you can't believe that it is the machine it says it is, which violates the most basic principles of the NCSC model. That's why electronic signatures are a good thing. I wrote (andy@sushi): > Then NCSC certification means nothing in many (most?) situations. Well, most sites are required to have certified systems (yet?). If they were, they wouldn't be allowed to have non-complying systems. The designers of the Ford Pinto were told by the US DOT to use $x as a cost-benefit tradeoff point for rear end collisions. Ford was still liable. I'd be surprised if NCSC certification protected a company from liability. (In other words, being right can be more important than complying.) [This case was cited again by Peter Browne (from old Ralph Nader materials?), at a Conference on Risk Analysis at NBS 15 September 1986: Ford estimated that the Pinto gas tank would take $11 each to fix in 400,000 cars, totalling $4.4M. They estimated 6 people might be killed as a result, at $400,000 each (the going rate for lawsuits at the time?), totalling $2.4M. PGN] Absolutely, network access should be as secure as phone access, IF YOU CHOOSE TO WORK IN THAT MODE. Our links to the outside world are as tightly restricted as our dialins. The Berkeley networking software is set up to support a much more integrated kind of network, where the network is treated as a single system. For our development environment that is much more effective. You should never allow that kind of access to a machine you don't control. Never. My interpretation of the original note was that the author's net contained machines with trusted-host access which should not have had such access; I contend that that represents NOT a failing of the software, but a failing of the administration of the network. My interpretation of Brian's original message is that he didn't have a choice; Berkeley network software trusts hosts on the local net. If that's true, then the administrators didn't have a chance to fail; the software's designers had done it for them. (I repeated all of Scott's paragraph because I agree with most of what he had to say.) -andy [I think the implications are clear. The network software is weak. Administrators are often unaware of the risks. Not all hosts are trustworthy. The world is full of exciting challenges for attackers. All sorts of unrealistic simplifying assumptions are generally made. Passwords are typically stored or transmitted in the clear and easily readable or obtained -- or else commonly known. Encryption is still vulnerable if the keys can be compromised (flawed key distribution, unprotected or subject to bribable couriers) or if the algorithm is weak. There are lots of equally devastating additional vulnerabilities waiting to be exercised, particularly in vanilla UNIX systems and networks thereof. Remember all of our previous discussions about not trying to put the blame in ONE PLACE. PGN] ------------------------------ RISKS-LIST: RISKS-FORUM Digest Thursday, 25 September 1986 Volume 3 : Issue 66 From: reid@decwrl.DEC.COM (Brian Reid) Date: 25 Sep 1986 0014-PDT (Thursday) To: risks@sri-csl.ARPA Reply-To: Reid@sonora.DEC.COM Subject: Follow-up on Stanford breakins: PLEASE LISTEN THIS TIME! "What experience and history teach is that people have never learned anything from history, or acted upon principles deduced from it." -- Georg Hegel, 1832 Since so many of you are throwing insults and sneers in my direction, I feel that I ought to respond. I am startled by how many of you did not understand my breakin message at all, and in your haste to condemn me for "asking for it" you completely misunderstood what I was telling you, and why. I'm going to be a bit wordy here, but I can justify it on two counts. First, I claim that this topic is absolutely central to the core purpose of RISKS (I will support that statement in a bit). Second, I would like to take another crack at making you understand what the problem is. I can't remember the names, but all of you people from military bases and secure installations who coughed about how it was a network administration failure are completely missing the point. This is a "risks of technology" issue, pure and simple. As an aside, I should say that I am not the system manager of any of the systems that was broken into, and that I do not control the actions of any of the users of any of the computers. Therefore under no possible explanation can this be "my fault". My role is that I helped to track the intruders down, and, more importantly, that I wrote about it. I am guessing that most of you are college graduates. That means that you once were at a college. Allow me to remind you that people do not need badges to get into buildings. There are not guards at the door. There are a large number of public buildings to which doors are not even locked. There is not a fence around the campus, and there are not guard dogs patrolling the perimeter. The university is an open, somewhat unregulated place whose purpose is the creation and exchange of ideas. Freedom is paramount. Not just academic freedom, but physical freedom. People must be able to walk where they need to walk, to see what they need to see, to touch what they need to touch. Obviously some parts of the university need to be protected from some people, so some of the doors will be locked. But the Stanford campus has 200 buildings on it, and I am free to walk into almost any of them any time that I want. More to the point, *you* are also free to walk into any of them. Now let us suppose that I am walking by the Linguistics building and I notice that there is a teenager taking books out of the building and putting them in his car, and that after I watch for a short while, I conclude that he is not the owner of the books. I will have no trouble convincing any policeman that the teenager is committing a crime. More important, if this teenager has had anything resembling a normal upbringing in our culture, I will have no trouble convincing the teenager that he is committing a crime. Part of the training that we receive as citizens in our society is a training in what is acceptable public behavior and what is not. The books were not locked up, the doors to the library were not locked, but in general people do not run in and steal all of the books. Or let me suppose instead that I am a reporter for the Daily News. I have a desk in a huge room full of desks. Most of the desks are empty because the other reporters are out on a story. You've seen scenes like this in the movies. It is rare in small towns to find those newsrooms locked. Here in Palo Alto I can walk out of my office, walk over to the offices of the Times Tribune a few blocks away, walk in to the newsroom, and sit down at any of those desks without being challenged or stopped. There is no guard at the door, and the door is not locked. There are 50,000 people in my city, and since I have lived here not one of them has walked into the newsroom and started destroying or stealing anything, even though it is not protected. Why not? Because the rules for correct behavior in our society, which are taught to every child, include the concept of private space, private property, and things that belong to other people. My 3-year-old daughter understands perfectly well that she is not to walk into neighbors' houses without ringing the doorbell first, though she doesn't quite understand why. People's training in correct social behavior is incredibly strong, even among "criminals". Murderers are not likely to be litterbugs. Just because somebody has violated one taboo does not mean that he will immediately and systematically break all of them. In some places, however, society breaks down and force must be used. In the Washington Square area of New York, for example, near NYU, you must lock everything or it will be stolen. At Guantanamo you must have guards or the Cubans will come take things. But in Palo Alto, and in Kansas and in Nebraska and Wisconsin and rural Delaware and in thousands of other places, you do not need to have guards and things do not get stolen. I'm not sure what people on military bases use computer networks for, but here in the research world we use computer networks as the building blocks of electronic communities, as the hallways of the electronic workplace. Many of us spend our time building network communities, and many of us spend our time developing the technology that we and others will use to build network communities. We are exploring, building, studying, and teaching in an electronic world. And naturally each of us builds an electronic community that mirrors the ordinary community that we live in. Networks in the Pentagon are built by people who are accustomed to seeing soldiers with guns standing in the hallway. Networks at Stanford are built by people who don't get out of bed until 6 in the evening and who ride unicycles in the hallways. Every now and then we get an intruder in our electronic world, and it surprises us because the intruder does not share our sense of societal responsibilities. Perhaps if Stanford were a military base we would simply shoot the intruder and be done with it, but that is not our way of doing things. We have two problems. One is immediate--how to stop him, and how to stop people like him. Another is very long-term: how to make him and his society understand that this is aberrant behavior. The result of all of this is that we cannot, with 1986 technology, build computer networks that are as free and open as our buildings, and therefore we cannot build the kind of electronic community that we would like. I promised you that I would justify what this all has to do with RISKS. We are developing technologies, and other people are using those technologies. Sometimes other people misuse them. Misuse of technology is one of the primary risks of that technology to society. When you are engineering something that will be used by the public, it is not good enough for you to engineer it so that if it is used properly it will not hurt anybody. You must also engineer it so that if it is used *improperly* it will not hurt anybody. I want to avoid arguments of just where the technologist's responsibility ends and the consumer's responsibility begins, but I want to convince you, even if you don't believe in the consumer protection movement, that there is a nonzero technologist's responsibility. Let us suppose, for example, that you discovered a new way to make screwdrivers, by making the handles out of plastic explosives, so that the screwdriver would work much better under some circumstances. In fact, these screwdrivers with the gelignite handles are so much better at putting in screws than any other screwdriver ever invented, that people buy them in droves. They have only one bug: if you ever forget that the handle is gelignite, and use the screwdriver to hit something with, it will explode and blow your hand off. You, the inventor of the screwdriver, moan each time you read a newspaper article about loss of limb, complaining that people shouldn't *do* that with your screwdrivers. Now suppose that you have invented a great new way to make computer networks, and that it is significantly more convenient than any other way of making computer networks. In fact, these networks are so fast and so convenient that everybody is buying them. They have only one bug: if you ever use the network to connect to an untrusted computer, and then if you also forget to delete the permissions after you have done this, then people will break into your computer and delete all of your files. When people complain about this, you say "don't connect to untrusted computers" or "remember to delete the files" or "fire anyone who does that". Dammit, it doesn't work that way. The world is full of people who care only about expediency, about getting their screws driven or their nets worked. In the heat of the moment, they are not going to remember the caveats. People never do. If the only computers were on military bases, you could forbid the practice and punish the offenders. But only about 0.1% of the computers are on military bases, so we need some solutions for the rest of us. Consider this scenario (a true story). Some guy in the Petroleum Engineering department buys a computer, gets a BSD license for it, and hires a Computer Science major to do some systems programming for him. The CS major hasn't taken the networks course yet and doesn't know the risks of breakins. The petroleum engineer doesn't know a network from a rubber chicken, and in desperation tells the CS student that he can do whatever he wants as long as the plots are done by Friday afternoon. The CS student needs to do some homework, and it is much more convenient for him to do his homework on the petroleum computer, so he does his homework there. Then he needs to copy it to the CS department computer, so he puts a permission file in his account on the CSD computer that will let him copy his homework from the petroleum engineering computer to the CSD computer. Now the CS student graduates and gets a job as a systems programmer for the Robotics department, and his systems programmer's account has lots of permissions. He has long since forgotten about the permissions file that he set up to move his homework last March. Meanwhile, somebody breaks into the petroleum engineering computer, because its owner is more interested in petroleum than in computers and doesn't really care what the guest password is. The somebody follows the permission links and breaks into the robotics computer and deletes things. Whose fault is this? Who is to blame? Who caused this breakin? Was it the network administrator, who "permitted" the creation of .rhosts files? Was it the person who, in a fit of expedience, created /usr/local/bin with 0776 protection? Was it the idiot at UCB who released 4.2BSD with /usr/spool/at having protection 0777? Was it the owner of the petroleum engineering computer? Was it the mother of the kid who did the breaking in, for failing to teach him to respect electronic private property? I'm not sure whose fault it is, but I know three things: 1) It isn't my fault (I wasn't there). It isn't the student's fault (he didn't know any better--what can you expect for $5.75/hour). It isn't the petroleum engineer's fault (NSF only gave him 65% of the grant money he asked for and he couldn't afford a full-time programmer). Maybe you could argue that it is the fault of the administrator of the CSD machine, but in fact there was no administrator of the CSD machine because he had quit to form a startup company. In fact, it is nobody's fault. 2) No solution involving authority, management, or administration will work in a network that crosses organization boundaries. 3) If people keep designing technologies that are both convenient and dangerous, and if they keep selling them to nonspecialists, then expedience will always win out over caution. Convenience always wins, except where it is specifically outlawed by authority. To me, this is one of the primary RISKs of any technology. What's special about computers is that the general public does not understand them well enough to evaluate the risks for itself. ------------------------------ End of RISKS-FORUM Digest ************************ ------- ---(127)---