A security flaw found in several of the most widely used e-mail programs (Microsoft Outlook Express, Microsoft Outlook 98, and Netscape Mail) could be used by malicious persons to send computers using those programs a virus that could destroy or steal data and could cause those computers to crash. The flaw, which is known as a buffer overflow error, occurs when a program fails to check the length of each character string. This failure means that a string too large to fit into an allotted memory location will lock up the program and fool the operating system into running attacker code in its place. Whereas new languages such as Java have built-in safeguards to prevent this kind of programmer error, older languages such as C and C++ do not. Computer security specialist Steven Bellovin says, "C makes it too easy to slice your fingers off, and programmers all over the world are doing so with great regularity." (*The New York Times*, 30 Jul 1998; Edupage, 30 July 1998. This is the Finnish find.)
There's been a lot of coverage of the potential security hole in Microsoft and Netscape's mail readers. Most reports note that this is the same kind of bug exploited by Robert Morris's Internet worm: overflow a fixed-length buffer on the stack to predictably sabotage the control flow of the program. What all the reports I've read appear to be missing is that bugs like this are almost inevitable in C or C++, yet pose almost no security issues in safer programming languages, including as Java, Lisp, Ada, Smalltalk, Modula-3, Eiffel, ML, etc. That is, the consequence of overflowing an array in Java is that an ArrayIndexOutOfBoundsException is thrown. No stack gets overwritten, no data gets corrupted. The program fails, but it does so in a sensible, predictable manner. Contrast this with C or C++, where the existence of a single, unchecked array access based on user-provided input is sufficient to expose gaping security holes. Just as the use of four digits for dates was considered wasteful or unnecessary in the past, the use of safe languages is often thought a luxury today. It isn't.
ISS Security Alert Advisory August 6th, 1998 Cult of the Dead Cow Back Orifice Backdoor Synopsis: A hacker group known as the Cult of the Dead Cow has released a Windows 95/98 backdoor named 'Back Orifice' (BO). Once installed this backdoor allows unauthorized users to execute privileged operations on the affected machine. Back Orifice leaves evidence of its existence and can be detected and removed. The communications protocol and encryption used by this backdoor has been broken by ISS X-Force. Description: A backdoor is a program that is designed to hide itself inside a target host in order to allow the installing user access to the system at a later time without using normal authorization or vulnerability exploitation. Functionality: The BO program is a backdoor designed for Windows 95/98. Once installed it allows anyone who knows the listening port number and BO password to remotely control the host. Intruders access the BO server using either a text or graphics based client. The server allows intruders to execute commands, list files, start silent services, share directories, upload and download files, manipulate the registry, kill processes, list processes, as well as other options. Encrypted Communications: All communications between backdoor client and the server use the User Datagram Protocol (UDP). All data sent between the client and server is encrypted, however it is trivial to decrypt the data sent. X-Force has been able to decrypt BO client requests without knowing the password and use the gathered data to generate a password that will work on the BO server. The way that BO encrypts its packets is to generate a 2 byte hash from the password, and use the hash as the encryption key. The first 8 bytes of all client request packets use the same string: "*!*QWTY?", thus it is very easy to brute force the entire 64k key space of the password hash and compare the result to the expected string. Once you know the correct hash value that will decrypt packets, it is possible to start generating and hashing random passwords to find a password that will work on the BO server. In our tests in the X-Force lab, this entire process takes only a few seconds, at most, on a Pentium-133 machine. With our tools we have been able to capture a BO request packet, find a password that will work on the BO server, and get the BO server to send a dialog message to warn the administrator and kill its own process. Determining if BO has been installed on your machine: The BO server will do several things as it installs itself on a target host: * Install a copy of the BO server in the system directory (c:\windows\system) either as " .exe" or a user specified file name. * Create a registry key under HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\RunServices with the file name of the server file name and a description field of either "(Default)" or a user specified description. * The server will begin listening on UDP port 31337, or a UDP port specified by the installer. You can configure RealSecure to monitor for network traffic on the default UDP 31337 port for possible warning signs. In order to determine if you are vulnerable: 1. Start the regedit program (c:\windows\regedit.exe). 2. Access the key HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\RunServices. Look for any services that may not have been intentionally installed on the machine. If the length of one of these file is close to 124,928 (give or take 30 bytes) then it is probably BO. Recommended action: BO can be removed by deleting the server and removing its registry entry. If possible, you should back up all user data, format your hard drive, and reinstall all operating systems and software on the infected machine. However, if someone has installed BO on your machine, then it is most likely part of a larger security breach. You should react according to your site security policy. Determining the password and configuration of an installed BO: 1. Using a text editor like notepad, view the server exe file. 2. If the last line of the file is '8 8$8(8,8084888<8@8D8H8L8P8T8X8\8'8d8h8l8', then the server is using the default configuration. Otherwise, the configuration will be the last several lines of this file, in this order: <filename> <service description> <port number> <password> <optional plugin information> Conclusion: Back Orifice provides an easy method for intruders to install a backdoor on a compromised machine. Back Orifice's authentication and encryption is weak, therefore an administrator can determine what activities and information is being sent via BO. Back Orifice can be detected and removed. This backdoor only works on Windows 95 and Windows 98 for now and not currently on Windows NT. [Copyright (c) 1998 by Internet Security Systems, Inc. Permission is hereby granted for the redistribution of this alert electronically. It is not to be edited in any way without express consent of X-Force. If you wish to reprint the whole or any part of this alert in any other medium excluding electronic medium, please e-mail firstname.lastname@example.org for permission.] Disclaimer The information within this paper may change without notice. Use of this information constitutes acceptance for use in an AS IS condition. There are NO warranties with regard to this information. In no event shall the author be liable for any damages whatsoever arising out of or in connection with the use or spread of this information. Any use of this information is at the user's own risk. X-Force Vulnerability and Threat Database: http://www.iss.net/xforce Please send suggestions, updates, and comments to: X-Force <email@example.com> of Internet Security Systems, Inc.
"More kindling for the fire: http://www.news.com/News/Item/0,4,24643,00.html Yet another serious NT security bug announced and acknowledged by Microsoft. A flaw in the OS would allow any ordinary network user to impersonate an NT system administrator. There is also the *possibility* that anyone with Internet access could do the same thing. How secure is national security? Let's hope that NT has not found its way aboard any SSBNs... Linux anyone? Mitch Stanek
I heard part of National Public Radio's Science Friday on July 24 called "Beyond Windows" (you can see more -- a list of the guests, for instance -- and hear the show at http://www.sciencefriday.com/pages/1998/Jul/hour1_072498.html). The part I heard was at the end of the show. Someone said that it's rumored Windows NT 5.0 will contain 50M lines of C code. Several comments followed: C was too primitive for the task, but C++ was too complex; 30M lines of the 50M are probably IF statements figuring out what environment it's running in and acting accordingly; isn't 50M lines of code something like what the Star Wars missile defense system was going to take and nobody thought that would ever work. Fred Ballard
The U.S. Department of Justice is now saying that it does not support the proposed amendments to the Communications Assistance for Law Enforcement Act (CALEA) that the FBI had provided to Senators a few weeks ago (See EPIC Alert 5.10). Assistant Attorney General Steven Colgate characterizes the amendment as a "staff document" and describes the language on emergency access to cell phone location information without a warrant as "boneheaded." However, Senate staff reports receiving calls from a senior FBI lobbyist pushing for the amendment even after the New York Times reported on the Bureau proposal. >From POLITECH -- the moderated mailing list of politics and technology To subscribe: send a message to firstname.lastname@example.org with this text: subscribe politech More information is at http://www.well.com/~declan/politech/
>From *Sydney Morning Herald*, 5 Aug 1998, p2: Electronic chaos as exchange overloads by Andrew Cassell [with JC additions from ABC radio] Bank computers crashed and data and fax lines were cut across the State yesterday when a power surge hit Telstra's Haymarket exchange in the city. The breakdown created extra queues at banks as computers, automatic teller machines and EFTPOS outlets failed. TAB [off course betting] agencies around the State were also affected and 3,500 Telstra customers in Mascot and 400 in Randwick lost phone services. The shutdown, between 12.20 pm and 1.20 pm, could have been much worse, Telstra's public affairs manager, Ms Kerrina Lawrence, said, but fast action by technicians had stopped the problem spreading through the whole of Sydney. [The technicians noticed the surge and moved isolate things manually. They then had to do 1000 manual resets. Mascot and Randwick are a few km from the affected exchange. Presumably, they didn't move to manual before the surge hit those exchanges as well.] Telstra call connect and 013 [Directory Assistance] services had also been disrupted, and in some cases, disconnected because of the surge, with the situation made worse by customers clogging up lines trying to get through. Problems with those services were continuing last night. [3 of the 5 largest Australian banks reported problems. Commonwealth Bank was affected until 5 pm.] John Colville: Centre for Teaching and Learning, Computing Sciences, +61-2-9514-2487 (1998) email@example.com
The following URL documents an interesting case of GPS interference. http://www.osl.com/ecdis/paper02.html Martin Poole, Perot Systems Europe firstname.lastname@example.org [The item is "Detrimental Effects of Installing Consumer Electronics on Ships" by Ken Hamer of Offshore Systems Ltd., Vancouver, presented at RTCM, May 1997. OSL makes ECDIS (Electronic Chart Display and Information System), based on a combination of GPS receiver and differential beacon receiver that enhances GPS accuracy -- in combination referred to as Differential GPS (DGPS). The article documents their efforts to identify the cause of a persistent erratic failure mode that occurred aboard the M.V. Manatoulin, a Great Lakes cargo ship. The narrative reads a little like a mystery story, and eventually leads to the captain's quarters' television antenna -- which used a cheap and noisy RF amplifier. PGN Stark Abstracting]
We seem to be too quick to blame WinNT for putting USS Yorktown dead in the water. The problem was described as someone entering a zero, after which a database overflowed and the propulsion system failed. That sounds like a bug in an application program, not the esteemed operating system! We're unlikely to learn more about the application itself. Visual A(da), Visual B(asic), Visual C. One could easily forget which language one is supposed to program with. This leads to the following thought - if a technophobe were intending to join the modern armed forces, which is the safest branch? It would certainly be less comfortable to have the ship handicapped by a systems failure in the submarine and air force services, and best in the Army where one could always bring up the backup systems: your feet. Andy Fraser
A simple solution: kill two birds with one stone by forcing Microsoft to GPL the source code to Windows NT. 1. we will finally be able to patch NT to make it secure and reliable enough for mission critical applications. 2. the software industry will be freed of an unfriendly tyrant, and Microsoft will be forced to compete in the office software arena on a freshly levelled playing field. "I am commandeering this source in the name of national security!" :-) Jonathan <email@example.com>
*Aviation Week and Space Technology*, 20 Jul 1998 Investigators believe two software errors and an improper command led to a loss of contact with the NASA/European Space Agency Solar and Heliospheric Observatory (SOHO) spacecraft on 24 Jun 1998. Recovery efforts are underway. An error in a preprogrammed command sequence resulted in an incorrect gyroscope reading, sending the spacecraft into an Emergency Sun Reacquisition (ESR) mode. A separate command sequence lacked code to activate a gyro needed for control when the spacecraft entered the ESR mode. Finally, a decision to command SOHO to turn off a gyro in response to unexpected telemetry caused the spacecraft to enter a series of ESRs, and ultimately led to loss of control, the agencies said. Nancy Leveson
On the lighter side of the Y2K problem, I recently received the following letter from the New Jersey Division of Motor Vehicles: Dear Motorist: Due to law enforcement concerns over the similarity of the green, year 2000 inspection stickers with the recently expired green stickers issued in 1997, the Division has discontinued use of the green sticker and is replacing it with a new orange, year 2000 sticker. Our records indicate that a year 2000 sticker was issued for your vehicle earlier this year ... We will replace your present sticker with a new orange one. At least we have the New Jersey government sticking it to us on year 2000 compliance. Still, even with a new orange Y2K sticker, I wonder if my car will be truly Y2K compliant. Dan Wallach Princeton University, CS Department firstname.lastname@example.org http://www.cs.princeton.edu/~dwallach/ [Crockwork Orange? PGN]
According to *The Guardian* (30 Jul 1998), Railtrack which was formed out of the old British Rail has no safety critical computer systems that need to be debugged. This is a legacy of underfunding over the past 20 years before it was privatized. The article quotes Railtrack as having 750 manual level signal boxes, 250 power boxes introduced in the 1980's and 9 electronically controlled boxes. It has therefore decided to downgrade it's preparation for the year 2000 as it rushes forward; out of the steam age. Mike Ellims, Pi Technology, email@example.com www.pi-group.com +44(0)1223441256
http://cgi.pathfinder.com/netly/0,2326,201980729-14220,00.html TIME.com / The Netly News, 29 Jul 1998 White House Calm, DoD Nervous About Y2K By Declan McCullagh (firstname.lastname@example.org) Few were surprised when John Koskinen, the White House's Y2K czar, said yesterday that "it's too early to say that in fact there are going to be major disruptions" due to the Year 2000 problem. Koskinen's work-hard-and-don't-be-scared advice is what the Clinton administration has been saying all along. But some of the Y2K experts Koskinen brought with him to a National Press Club briefing yesterday offered some dismaying details. Usually if, say, a 4,000-megawatt power plant gives up the ghost, it's no big deal. The electric industry is pretty good at planning for these sorts of breakdowns. But if dozens crash within a few hours on 1-1-00? "It's a very complex system," admitted Michael Gent, president of the North American Electric Reliability Council. "It's probably the most complex system every invented by man, more complicated than a moon shot." Gent nevertheless predicted that even if today were December 31, 1999, "the lights would stay on in most places." If they go out, it'll be the fault of the private sector, not the feds, Koskinen said. Contradicting the now-popular belief that the federal government's computers are in the worst shape, he predicted that "the threats to the economy and the public are not going to be federal systems." [...] Already prepared to accept his part is John Hamre, deputy secretary of defense. "I think we're probably going to be the poster child for failure," he said last week during a speech to Fortune 500 executives. "Nobody cares if the Park Services computers don't come on. OK? But what's going to happen if some do[n't] in the DoD?"
There is a good article on the html "referer" field in the magazine Web Techniques, Sept. 1998, page 10 - security risks by leaking information even from ssl pages or from behind firewalls, the right and wrong ways to use this field, and a problem that some sites had because of a change in implementing this field in Microsoft IE4 (illustrating the risk of using a field in a way it was not intended for).
Anyone looking for a book that puts many of the "software is speech" issues into sharp focus should get a copy of "Cracking DES: Secrets of Encryption Research, Wiretap Politics & Chip Design" (ISBN 1-56592-520-3). This book describes the DES-cracking machine that was recently in the news when it broke a 56-bit DES key in less than three days. The bulk of the book is also available from John Young's website, <http://jya.com/> -- and legally so, as the book, with a few small exceptions, is explicitly in the public domain. It's also available from the usual web-based booksellers. You'll find a substantial discussion of the politics of encryption research, the history of DES cracking, as well as all the source code and hardware design documents. This includes well-commented PERL and C source code (software source code intended to be read), the VHDL code that defines the custom chips used in the cracking engine, hardware board schematics and design notes. The source code is published using checksum tools that greatly simplify the task of scanning the book to recover the original text. While non-technical readers would not be expected to understand the computer source code and/or hardware design, the architecture overview and the various "political" sections should be well worth reading and within the grasp of anyone who can get through a Scientific American article without falling asleep. Martin Minow, email@example.com [Note: John Young's archive of the (legal to read) scanned volume is at <http://www.jya.com/cracking-des.htm>. Also, chapters that would be illegal to put on the Internet in the U.S. are available in abroad <ftp://ftp.nic.surfnet.nl/surfnet/net-security/encryption/cracking_DES/> .]
SEI Software Engineering Symposium 14-17 September 1998 David L. Lawrence Convention Center Pittsburgh, Pennsylvania Theme: Improving What You Build Means Improving How You Build Deadline for early bird registration discount: August 12, 1998. For complete details, visit our Web site at: http://www.sei.cmu.edu/products/events/symp/ For More Information about the Software Engineering Institute or for more details about the Software Engineering Symposium and other events, please contact SEI Customer Relations, 1-412-268-5800, firstname.lastname@example.org
Please report problems with the web pages to the maintainer