If you feel that you may have found a new virus, or are not quite sure if some file or boot sector is infected, please refer to Section F Question #4 (F4) before posting a request for assistance. The answer to this question has been developed to ensure new readers of Virus-L/comp.virus understand the protocol for raising such questions and to help them avoid asking questions that can be answered in this document. If you are looking for help in designing and implementing an antivirus policy or system, read all of Sections B through F inclusive, paying particular attention to Section D.
Please read the full list of questions carefully--as with most complex topics, dozens of different virus-related questions turn out to be about similar phenomena. If you don't find your exact question here, look closely at the ones that seem vaguely similar.
Above all, remember that the time to really worry about viruses is before your computer gets one!
The one-piece FAQ sheet should be available in a file called vlfaqxyy.txt, where "xyy" is the current version number (starting from 200 in mid-1995 for version 2.00). The multi-part version is created by splitting the main FAQ sheet into four pieces as follows:
vlfxyy-1.txt contains FAQ sections A & B
vlfxyy-2.txt contains FAQ sections C & D
vlfxyy-3.txt contains FAQ sections E & F
vlfxyy-4.txt contains FAQ sections G
(with "xyy" again representing the current version number). Please do not make your own multi-part FAQ, as each of the parts in the "official" multi-part version include additional preface information.
Either or both versions may also be available in some form of compressed archive--in this case the "name part" of the filename should be the same as the original file with the extension being replaced (or appended) as appropriate for the archiving method used. Please do not repackage the multi-part FAQ into one large archive file, as this defeats the sole purpose for creating it--to ensure that the FAQ sheet is "officially" available in a readable form that will pass unmolested through most e-mail gateways.
All the files in either version of the FAQ sheet are signed with Nick FitzGerald's PGP key. Nick's public key can be retrieved from the main PGP key servers. If you do not know what PGP is, but wish to validate your copy of the FAQ sheet, you should read the USENET newsgroup alt.security.pgp [please do not e-mail me, as I am not a PGP expert--FAQ maintainer].
The FAQ sheet is a dynamic document, changing as people's questions change. The version number also changes as any changes are made. Version numbers containing a "d" are drafts and should not be made publicly available, nor distributed. We ask for your cooperation in deleting and not further distributing "d" versions of the FAQ sheet. If you have any questions or contributions, please e-mail them to the FAQ sheet maintainer, Nick FitzGerald, at: n.fitzgerald@csc.canterbury.ac.nz
The most recent copy of the FAQ sheet will always be available on the Virus-L/comp.virus archives, including by anonymous FTP on corsa.ucr.edu (IP = 138.23.166.133) in the directory pub/virus-l.
A WWW version of the FAQ sheet, with cross-references and file links is currently under development, as is a WinHelp version with cross-references (if you would like to assist with these efforts, or to port one of these formats to another popular hypertext help format, please contact the FAQ sheet maintainer so we can better coordinate this work).
In various places the FAQ sheet mentions products by name. This is usually only for illustrative purposes. Such references should not be taken to imply that all, some, or any of the contributors to this FAQ sheet endorse any such product for any purpose or that such products are the best examples of what is being discussed. Such refernces are usually because the products named were the first to implement a particular feature or function. Further, that a given product is not mentioned in the FAQ should not be taken as an indication of its quality or suitability for any task.
Various brand and product names are used throughout the FAQ sheet--these remain trademarks or registered trademarks of their respective holders.
Unless indicated otherwise, prices are given in US dollars and should be taken as guides only. Telephone numbers include an indication of the time-zone relative to GMT--some of these are very approximate, but should be close enough to save you ringing in the middle of the receiver's night!
SUB VIRUS-L Jane DoeTo be removed from the Virus-L mailing list, send a message to LISTSERV@LEHIGH.EDU saying "SIGNOFF VIRUS-L".
To "subscribe" to comp.virus, simply use your favorite USENET news reader to read the group.
There are several major sources of information on specific viruses. Probably the largest one is Patricia Hoffman's hypertext VSUM. While VSUM is quite complete it only covers PC viruses and it is regarded by many in the antivirus field as being inaccurate, so we advise you not to rely solely on it. It can be downloaded from most major archive sites.
A more precise source of information is the Computer Virus Catalog, published by the Virus Test Center in Hamburg. It contains highly technical descriptions of computer viruses for several platforms: DOS, Mac, Amiga, Atari ST and Unix. Unfortunately, the DOS section is quite incomplete. The CVC is available by anonymous FTP from ftp.informatik.uni-hamburg.de (IP = 134.100.4.42), directory pub/virus/texts/catalog. (A copy of the CVC is also available by anonymous FTP on corsa.ucr.edu in the directory pub/virus-l/docs/vtc.)
Another small collection of good technical descriptions of PC viruses, called CARObase is also available from ftp.informatik.uni-hamburg.de, in the directory /pub/virus/texts/carobase.
A fourth source of information is the monthly Virus Bulletin, published in the UK. Among other things, it gives detailed technical information on viruses (see A8); a one year subscription, however, costs $395. US subscriptions can be ordered by calling (203) 431 8720 (GMT-5/-4) or writing to 590 Danbury Road, Ridgefield, CT 06877; for European subscriptions, the number is +44 1235 555139 (GMT+0/-1) and the address is: 21 The Quadrant, Abingdon, OXON, OX14 3YS, ENGLAND. General enquiries can be sent to virusbtn@vax.ox.ac.uk.
Another source of information is the book "Virus Encyclopedia" which is part of the printed documentation of Dr. Solomon's AntiVirus ToolKit (a commercial DOS antivirus program). It is more complete than the CVC list and just as accurate; however it lists only DOS viruses. This book may be available separately
The on-line help system of the shareware antivirus product Anti-Virus Pro contains a large and relatively exact collection of virus descriptions and even includes demonstrations of several of the audio and visual effects produced by some viruses. However the text can be difficult to read because English is not the author's native tongue.
The WWW site www.datafellows.fi has an on-line, cross-referenced database containing descriptions of about 1500 PC viruses, with an emphasis on viruses "in the wild". Another network-accessible source of information pertaining to viruses is provided by IBM AntiVirus, at A HREF="http://www.brs.ibm.com/ibmav.html">http://www.brs.ibm.com/ibmav.html or via gopher at the site index.almaden.ibm.com (choose "IBM Computer Virus Information Center" from the main menu).
An excellent source of information regarding Apple Macintosh viruses is the on-line documentation in the freeware Disinfectant program by John Norstad of Northwestern University. This is available at most Mac archive sites.
Many freeware/shareware DOS antivirus programs are available from the SimTel Software Repository. This collection of software is available via anonymous FTP from ftp.coast.net (IP = 141.210.10.117), with antivirus software in the directory /SimTel/msdos/virus. Note that the SimTel archive is "mirrored" at many anonymous FTP sites, including wuarchive.wustl.edu (IP = 128.252.135.4, /systems/ibmpc/simtel/virus), and nic.funet.fi (IP = 128.214.248.6, /pub/msdos/SimTel/virus). Most of this software can also be obtained via e-mail in uuencoded form from various TRICKLE sites, especially in Europe.
Likewise, Macintosh antivirus programs can be found in /pub/tools/mac at coast.cs.purdue.edu.
A list of many antivirus programs, including commercial products and one person's rating of them, can be obtained by anonymous ftp from corsa.ucr.edu (IP = 138.23.166.33) in pub/virus-l/docs/reviews in the file slade.quickref.rvw. This directory also contains detailed product reviews of many products.
Computers Under Attack: Intruders, Worms and Viruses edited by Peter J. Denning, ACM Press/Addison-Wesley, 1990. This is a book of collected readings that discuss computer viruses, computer worms, break-ins, and social aspects, and many other items related to computer security and malicious software. A very solid, readable collection that doesn't require a highly-technical background. Price: $20.50.
Rogue Programs: Viruses, Worms and Trojan Horses edited by Lance J. Hoffman, Van Nostrand Reinhold, 1990. This is a book of collected readings describing in detail how viruses work, where they come from, what they do, etc. It also has material on worms, Trojan Horse programs, and other malicious software programs. This book focuses more on mechanism and relatively less on social aspects than does the Denning book; however, there is an excellent piece by Anne Branscomb that covers legal aspects. Price: $32.95.
A Pathology of Computer Viruses by David Ferbrache, Springer-Verlag, 1992. This is an in-depth book on the history, operation, and effects of computer viruses. It is one of the most complete books on the subject, with an extensive history section, a section on Macintosh viruses, network worms, and Unix viruses. Price $49.00.
A Short Course on Computer Viruses, 2nd edition, by Dr. Fred B. Cohen, Wiley, 1994. This book is by a well-known pioneer in virus research, who has also written dozens of technical papers on the subject. Price: $35.00 ($45.00 with accompanying diskette).
Robert Slade's Guide to Computer Viruses, by Robert Slade, Springer-Verlag, 1994. This book is a comprehensive introduction to computer viruses, written in a clear and easy style for non-experts. Price $29.00.
A somewhat dated, but still useful, high-level description of viruses, suitable for a complete novice with little computer background is Computer Viruses: Dealing with Electronic Vandalism and Programmed Threats by Eugene H. Spafford, Kathleen A. Heaphy, and David J. Ferbrache, ITAA (Arlington, VA), 1989. ITAA (Information Technology Association of America) is a computer industry service organization and not a publisher. While many people have indicated they find this a very understandable reference it is now out of print, but portions of it have been reprinted in many other places, including Denning and Hoffman's books (above).
It is also worth consulting various publications such as Computers & Security and SECURE Computing (both of which, while not limited to viruses, contain many relevant papers) and the Virus Bulletin (published in the UK, it contains many technical articles).
There are several reasons for the apparent imbalance. One very general reason is that users of DOS heavily outnumber the users of other operating systems. The discussion in Virus-L/comp.virus therefore tends to have a preponderance of questions and chat about DOS specific infections and problems. We welcome questions, comments and reports from users of other operating systems and platforms. If you use a computer of another type, please do contribute to the discussion. Just because the majority are talking about DOS does not mean that your contribution is not welcome. It may be important precisely because you have a different perspective.
Therefore, let us assure you there is no deliberate attempt being made to exclude Amiga, Atari, Macintosh, OS/2, UNIX, VMS, Windows (NT, '95 or any other flavor) or any other platform or operating system from the discussion or the FAQ sheet. If you feel that there is too much PC bias, please don't complain about it--tell us something about the virus situation on your system.
The problem with Cohen's human language definition is that it doesn't capture many of the subtleties of his mathematical model--as indeed, few informal definitions do--and questions arise that can only be answered by checking his formal model. Using his formal definitions, Cohen classifies some things as viruses that most readers of Virus-L/comp.virus (and many experts) would not consider viruses. For example, given certain circumstances on an IBM PC running DOS, the DISKCOPY program is classified as a virus by Cohen's formalisms.
This has led to some tension between what Cohen considers a "virus" and what is usually discussed on Virus-L. Several other definitions of "virus" have been proposed, but it is probably fair to say that most of us are concerned about things that are viruses by the following definition:
A computer virus is a self-replicating program containing code that explicitly copies itself and that can "infect" other programs by modifying them or their environment such that a call to an infected program implies a call to a possibly evolved copy of the virus.Probably the major distinction between Cohen's definition and "viruses" as we tend to use the word is that we see them as deliberately designed to replicate (although there is some debate over this too). Cohen's definition does not require this (and this would be difficult to build into his formal model).
Note that many people use the term "virus" loosely to cover any sort of program that tries to hide its possibly malicious function and\or tries to spread onto as many computers as possible, though some of these programs may more correctly be called "worms" (see B2) or "Trojan Horses" (see B3). Also be aware that what constitutes a "program" for a virus to infect may include a lot more than is at first obvious--don't assume too much about what a virus can or can't do!
These software "pranks" are very serious; they are spreading faster than they are being stopped, and even the least harmful of viruses could be life-threatening. For example, in the context of a hospital life-support system, a virus that "simply" stops a computer and displays a message until a key is pressed, could be fatal. Further, those who create viruses can not halt their spread, even if they wanted to. It requires a concerted effort from computer users to be "virus-aware", rather than continuing the ambivalence that has allowed computer viruses to become such a problem.
Computer viruses are actually a special case of something known as "malicious logic" or "malware", and other forms of malicious logic are also discussed in Virus-L/comp.virus. It can be important to understand the distinctions between viruses and these other forms of malware.
Note that unlike viruses, worms do not need to attach themselves to a host program. There are two types of worms--host computer worms and network worms.
Host computer worms are entirely contained in the computer they run on and use network connections only to copy themselves to other computers. Host computer worms where the original terminates itself after launching a copy on another host (so there is only one copy of the worm running somewhere on the network at any given moment), are sometimes called "rabbits."
Network worms consist of multiple parts (called "segments"), each running on different machines (and possibly performing different actions) and using the network for several communication purposes. Propagating a segment from one machine to another is only one of those purposes. Network worms that have one main segment which coordinates the work of the other segments are sometimes called "octopuses."
The infamous Internet Worm (perhaps covered best in "The Internet Worm Program: An Analysis," Eugene H. Spafford, Purdue Technical Report CSD-TR-823) was a host computer worm, while the Xerox PARC worms were network worms (a good starting point for these is "The Worm Programs--Early Experience with a Distributed Computation," Communications of the ACM, 25, no.3, March 1982, pp. 172-180).
File infectors can be either DIRECT-ACTION or RESIDENT. A direct-action virus selects one or more programs to infect each time a program infected by it is executed. A resident virus installs itself somewhere in memory (RAM) the first time an infected program is executed, and thereafter infects other programs when they are executed (as in the case of the Jerusalem virus) or when other conditions are fulfilled. Direct-action viruses are also sometimes referred to as NON-RESIDENT. The Vienna virus is an example of a direct-action virus. Most viruses are resident.
The second main category of viruses is SYSTEM or BOOT-RECORD INFECTORS: these viruses infect executable code found in certain system areas on a disk. On PCs there are ordinary boot-sector viruses, which infect only the DOS boot sector, and MBR viruses which infect the Master Boot Record on fixed disks and the DOS boot sector on diskettes. Examples include Brain, Stoned, Empire, Azusa and Michelangelo. All common boot sector and MBR viruses are memory resident.
To confuse this classification somewhat, a few viruses are able to infect both files and boot sectors (the Tequila virus is one example). These are often called "MULTI-PARTITE" viruses, though there has been criticism of this name; another name is "BOOT-AND-FILE" virus.
Aside from the two main classes described above, many antivirus researchers distinguish either or both of the following as distinct classes of virus:
FILE SYSTEM or CLUSTER viruses (e.g. Dir-II) are those that modify directory table entries so that the virus is loaded and executed before the desired program is. The program itself is not physically altered, only the directory entry of the program file is. Some consider these to be a third category of viruses, while others consider them to be a sub-category of the file infectors. LINK virus is another term occasionally used for these viruses, though it should be avoided, as "link virus" is commonly used in the Amiga world to mean "file infecting virus."
KERNEL viruses target specific features of the programs that contain the "core" (or "kernel") of an operating system (3APA3A is a DOS kernel virus and is also multipartite). A file infecting virus that can infect kernel program files is not a kernel virus--this term is reserved for describing viruses that utilize some special feature of kernel files (such as their physical location on disk or a special loading or calling convention).
Example: The very first DOS virus, Brain, a boot-sector infector, monitors physical disk I/O and re-directs any attempt to read a Brain-infected boot sector to the disk area where the original boot sector is stored. The next viruses to use this technique were the file infectors Number of the Beast and Frodo (aka 4096, 4K).
Countermeasures: A "clean" system is needed so that no virus is present to distort the results of system status checks. Thus the system should be started from a trusted, clean, bootable diskette before any virus-checking is attempted; this is "The Golden Rule of the Trade" (see G8 for help with making a clean boot disk and booting clean).
One method of evading scan string-driven virus detectors is self-encryption with a variable key. These viruses (e.g. Cascade) are not termed "polymorphic", as their decryption code is always the same. Therefore the decryptor can be used as a scan string by the simplest scan string-driven virus scanners (unless another virus uses the identical decryption routine and exact identification (see B15) is required).
A technique for making a polymorphic virus is to choose among a variety of different encryption schemes requiring different decryption routines: only one of these routines would be plainly visible in any instance of the virus (e.g. the Whale virus). A scan string-driven virus scanner would have to exploit several scan strings (one for each possible decryption method) to reliably identify a virus of this kind.
More sophisticated polymorphic viruses (e.g. V2P6) vary the sequences of instructions in their variants by interspersing the decryption instructions with "noise" instructions (e.g. a No Operation instruction or an instruction to load a currently unused register with an arbitrary value), by interchanging mutually independent instructions, or even by using various instruction sequences with identical net effects (e.g. Subtract A from A, and Move 0 to A). A simple-minded, scan string-based virus scanner would not be able to reliably identify all variants of this sort of virus; rather, a sophisticated "scanning engine" has to be constructed after thorough research into the particular virus.
One of the most sophisticated forms of polymorphism used so far is the "Mutation Engine" (MtE) which comes in the form of an object module. With the Mutation Engine any virus can be made polymorphic by adding certain calls to its assembler source code and linking to the mutation-engine and random-number generator modules.
The advent of polymorphic viruses has rendered virus-scanning an ever more difficult and expensive endeavor; adding more and more scan strings to simple scanners will not adequately deal with these viruses.
A FAST infector is a virus that, when it is active in memory, infects not only programs which are executed, but even those that are merely opened. The result is that if such a virus is in memory, running a scanner or integrity checker can result in all (or at least many) programs becoming infected. Examples are the Dark Avenger and the Frodo viruses.
The term "SLOW infector" is sometimes used to refer to a virus that only infect files as they are modified or as they are created. The purpose is to fool people who use integrity checkers into thinking that modifications reported by their integrity checker are due solely to legitimate reasons. An example is the Darth Vader virus.
Such a possibility however, need not translate into much of a threat. It is rare for modern software to require the computer it runs on to have an ANSI console, so few PCs or other machines should load ANSI drivers. Also, few people use software that simply "types" output to the terminal device, so such an ANSI bomb in an e-mail or News posting would most likely not reprogram your keyboard anyway. Further, although FORMAT C: may be catastrophic under certain versions of DOS, it won't hurt Macintoshes and would probably have very unexpected, or no, effects on other systems.
If you are at all worried about the possibility of having something untoward happen on your PC due to an ANSI bomb and you have to load an ANSI driver (some communications software still requires it), look for one of the third-party ANSI drivers which abound on BBSes and FTP sites. Most of these have improved performance over DOS's ANSI.SYS and either do not support, or let you disable, keyboard re-mapping.
BSI = Boot Sector Infector: a virus that takes control when the computer attempts to boot. These are found in the boot sectors of floppy disks, and the MBRs or boot sectors of hard disks (see B4 for more details). BSIs are also known as BSVs (Boot Sector Viruses).
CMOS = Complementary Metal Oxide Semiconductor: A memory area that is used in AT class, and higher, PCs for storage of system information. CMOS is battery backed RAM (see below), originally used to maintain date and time information while the PC was turned off. CMOS memory is not in the normal CPU address space and cannot be executed (see E2 for further discussion of issues concerning CMOS memory and viruses).
DBS = DOS Boot Sector: The first sector of a logical DOS partition on a hard disk or the first absolute sector of a diskette. This sector contains the startup code that actually loads DOS. This is often confused with the MBR. Some boot sector viruses infect the DBS rather than the MBR when infecting hard disks.
DETECTION = The ability of an antivirus program to detect that a virus is present, without necessarily reporting which particular virus it is (also see IDENTIFICATION and RECOGNITION, in this section).
DOS = Disk Operating System. We use the term "DOS" to mean any of the MS-DOS, PC-DOS, DR DOS or Novell DOS systems for PCs and compatibles, even though there are operating systems called "DOS" on other, unrelated machines.
GERM = The first generation of a virus. It normally cannot be produced again during the replication process and is usually created by compiling the source of the virus.
GOAT FILES = Programs which usually do nothing special (e.g., just exit, or simply display a message), that are used by antivirus researchers to capture samples of viruses. This is done to make it easier to disassemble and understand the virus, because the infected "goat" program is (usually) simple and does not clutter the disassembly. Alternative terms are BAIT FILES, DECOY FILES and VICTIM FILES. In any of these terms, the word "programs" often replaces the word "files".
IDENTIFICATION = The ability of an antivirus program (usually a scanner) to not only detect the virus and recognize it by name, but also to recognize it to a high degree of uniqueness. This allows third parties to understand which particular virus it is without seeing a sample of the virus. EXACT IDENTIFICATION occurs when every section of the non-modifiable parts of the virus body are uniquely identified. ALMOST EXACT IDENTIFICATION occurs if the identification is only good enough to ensure that an attempt to remove the virus will not result in damage to the host object by the use of an inappropriate disinfection method (also see DETECTION and RECOGNITION, in this section).
MBR = Master Boot Record: the first absolute sector (track 0, head 0, sector 1) on a PC hard disk, that usually contains the partition table but on some PCs may only contain a boot sector. The MBR is also known as the MBS (Master Boot Sector). This is not the same as the DOS Boot Sector, logical sector 0 (see above).
PARTITION TABLE = A 64-byte data structure that defines the way a PC's hard disk is divided into logical sections known as partitions. While there is often more than one partition table on a PC's hard disk, the most important is the one stored in the MBR. This one contains important extra information such as which partition (if any) should be booted from. The partition table is purely data, so is not executed. Some people erroneously use the term "partition table virus" as a synonym for "MBR virus".
RAM = Random Access Memory: the place programs are loaded into in order to execute; the significance for viruses is that, to be active, they must load themselves into part of the RAM. However, some virus scanners may declare that a virus is active when it is found in RAM, even though it may only be left in a buffer area following a disk read operation, rather than truly being active (see C8 for further discussion of this issue).
RECOGNITION = The ability of an antivirus program (usually a scanner) to detect a virus and to recognize it by name (also see DETECTION and IDENTIFICATION, in this section).
TARGETING VIRUS = A virus that tries to bypass or hinder the operation of one or more specific antivirus programs. Also known as RETALIATOR, RETRO and ANTI-ANTIVIRUS viruses.
SCAN STRING = A sequence of bytes (characters) that occur in a known virus but not, one hopes, in legitimate programs. Some scanners allow "wildcards"--positions that are matched by any character--in their scan strings. Authors of virus scanners reduce the likelihood of false positives (see C5) by carefully selecting their scan strings and often by only searching "likely" parts of target files.
SEARCH STRING = A synonym for scan string.
SIGNATURE = A poor synonym for scan string. We recommend that you avoid using this term and use "scan string" or "search string" instead.
TOM = Top Of Memory: the end of conventional memory--an architectural design limit--at the 640KB mark on most PCs. Some early PCs may not have a full 640KB, but the amount of memory is always a multiple of 64KB. A boot-record virus on a PC typically resides just below this mark and changes the value which will be reported for the TOM to the location of the beginning of the virus so that it won't be overwritten. Checking this value for changes can help detect a virus, but there are also legitimate reasons why it may change (see C10). A very few PCs with unusual configurations or memory managers may report in excess of 640KB.
TSR = Terminate but Stay Resident: these are PC programs that stay in memory while you continue to use the computer for other purposes; they include pop-up utilities, network software, and the great majority of common viruses. These can often be seen using utilities such as MEM and MSD.
VX = Virus eXchange. A shorthand usually reserved for those BBSes and FTP sites, and their community of users, that make their virus collections "openly" available for downloading. Exchange of virus samples between bona fide members of the antivirus community is not tagged with the VX label.
Thus, the more successful viruses typically try to spread as much as possible before delivering their payload, if any. As these tend to be the viruses you are most likely to encounter, you should be aware that there are usually symptoms of virus infection before any (or much!) damage is done.
There are various kinds of symptoms that some virus authors have written into their programs, such as messages, music and graphical displays. The main indications, however, are changes in file sizes and contents, changing of interrupt vectors, or the reassignment of other system resources. The unaccounted use of RAM or a reduction in the amount reported to be in the machine are important indicators. Examination of program code is valuable to the trained eye, but even a novice can often spot the gross differences between a valid boot sector and some viral ones. These symptoms, along with longer disk activity and strange behavior from the hardware, may instead be caused by genuine software, by harmless "joke" programs, or by hardware faults.
The only foolproof way to determine that a virus is present is for an expert to analyse the assembly code contained in all programs and system areas, but this is usually impracticable. Virus scanners go some way towards performing this analysis by looking in that code for known viruses; some even use heuristic means to spot "virus-like" code, but this is not always reliable. It is wise to arm yourself with the latest antivirus software and to pay close attention to your system. In particular, look for any unexpected change in the memory map or configuration as soon as you start the computer. For users of DOS 5.0+, the MEM program with the /C switch is very handy for this. If you have DR DOS, use MEM with the /A switch; if you have an earlier DOS version, use CHKDSK or the commonly-available MAPMEM utility. You don't have to know what all the numbers mean, only that they have changed unexpectedly. Mac users have "info" options, which give some indication of memory use, but may need ResEdit to supply more detailed information.
If you run Windows on your PC and you suddenly start getting messages at Windows startup that 32-bit Disk Access cannot be used, this often indicates your PC has been infected by a boot-sector virus.
If you run into an alarm and your scanner doesn't identify anything or doesn't properly clean up for you, first verify that the version you are using is the most recent. Then get in touch with a reputable antivirus researcher, who may ask you to send in a copy of the infected file. (Also see C9 and F4 if you decide you need to ask for help on Virus-L/comp.virus.)
If backups of infected or damaged files are available and, in making them, appropriate care was taken to ensure that infected files have not been included in the backups (see D10), restoring from backup is the safest solution, even though it can be a lot of work if many files are involved.
More commonly, a disinfecting program is used, though disinfection is somewhat controversial and problematic (see E8). If the virus is a boot-sector infector, you can continue using the computer with relative safety (if the hard disk's partition table is left intact) by booting from a clean system diskette. However, it is wise to go through all your diskettes removing any infections as, sooner or later, you will be careless and leave an infected diskette in the machine when it reboots, or give an infected diskette to a someone who doesn't have appropriate defenses to avoid infection.
Most PC boot-sector infections can be cured by the following simple process--pay particular care to make the checks in Steps 2 and 3.
Note that removing an MBR virus in the following way may not be desirable, and may even cause valuable information to be lost. For instance, the One_Half virus gradually encrypts the infected hard drive "inwards" (starting from the "end" and moving towards the beginning), encrypting two more tracks at each boot. The information about the size of the encrypted area is only stored in the MBR. If the virus is removed using the method above, this information will be irrecoverably lost and part of the disk with unknown size will remain encrypted.
You should continue with Step 4 only if all the tests in Step 2 and this step pass. Do NOT continue if you were unable to correctly access all your hard disks, as you will quite possibly damage critical information making permanent data damage or loss more likely.
For example, the Stoned virus replaces the disk's boot record with its own, relocating the original to a sector on the disk that may (or may not) occur in an unused portion of the root directory of a DOS diskette; when active, it sits in an area a few kilobytes below the top of memory. All this description could apply to a number of common viruses; but the important points of where the original boot sector goes--and what effect that has on networking software, non-DOS partitions, and so on--are all major questions in themselves.
Therefore, it is better if you first try to answer your question yourself. There are several sources of information about the known computer viruses, so please consult one of them before requesting information publicly. Chances are that your virus is rather well known and that it is already described in detail in at least one of these sources (see A6 for some help in finding these.)
Following from some of Fred Cohen's work, it has been proven that every virus detector must have an infinite number of false positives, false negatives, or both. This is expressed by saying that detection of viruses, either by appearance or behavior, is UNDECIDABLE. The interpretation and practical significance of this depends upon the interpretation of the terms used, and as with Fred's definition of the term "computer virus", there is some debate over this.
In the case of virus scanners, false positives are rare, but they can arise if the scan string chosen for a given virus is also present in some benign objects because the string was not well chosen. In modern scanners, most false positives probably occur because some virus encryption engines produce very "normal looking" code and scanners that only try to decide if a piece of code could have been generated by a known virus encryption procedure will occasionally detect "innocent" code as "suspicious". False negatives are more common with virus scanners because scanners will miss completely new or heavily modified viruses.
One other serious problem could occur: A positive that is misdiagnosed. As an example, imagine a scanner faced with the Empire virus in a boot record that reports it as the Stoned virus. In this case, use of a Stoned-specific "cure" to recover from an Empire infection could result in an unreadable disk or loss of extended partitions. Similarly, sometimes "generic" disinfection (see D1) can result in unusable files, unless a check is made (e.g. by comparing checksums) that the recovered file is identical to the original file. The better generic disinfection products all store information about the original files to allow verification of recovery processes.
A particular type of false positive, where (part of) an inactive virus is detected, is known as a GHOST POSITIVE. Ghost positives usually occur in one of four situations (the first two of which are examples of antivirus programs "upsetting" each other):
Ghost positives can be caused when the disinfection routine of an antivirus program "unhooks" a virus from its target (be it a file or boot sector) but it does so in such a way that part of the virus code is left intact (though that code will never be executed). Another antivirus program might see this code and report it is an infection. In this case the second antivirus program is seeing a "ghost"--part of a virus that was there.
A scanner may "see" the unencoded scan strings of another scanner, left in memory after the first has run or held in memory by a resident scanner, and report these "ghosts" as active viruses (see C6 and C8).
As explained elsewhere (see E10) a copy of an infected diskette boot sector, sitting in the disk buffers, may be detected and reported as an active virus.
Disinfection procedures can result in virus "remnants" being left in "slack space" (disk space allocated to files but not actually occupied). As in the case of copies of infected diskette boot sectors being held in disk buffers, these remnants can be detected and incorrectly reported as being active. Ghost positives of this nature should disappear after running disk defragmentation or "optimization" programs with the option to "clean" slack space. Occasionally running a defragmenter (like MS-DOS 6's DEFRAG) after a full data backup (see D10), is a good idea anyway--especially before installing new software. Unfortunately, DOS's DEFRAG does not have a "clean slack space" option, though some third-party defragmenters do. There are also utilities that clean unallocated and slack space and these should remove ghost positives caused by "remnants".
Most antivirus programs try very hard to identify viral infections only, but sometimes they give false alarms (see C5). If two different antivirus programs are both of the "scanner" type, they will contain "scan strings" from which they identify viral infections. If the strings are not "encoded", then they may be identified as a virus by another scanner type program. Also, if the scanner does not remove the strings from memory after it has run, then another scanner may detect a virus string "in memory". This often causes the second scanner to report that your system is "infected", but only after you have run the first scanner (which may be a memory resident one). The major contributors to this group are so tired of dealing with non-virus reports of this sort that they strongly recommend users to avoid antivirus software which doesn't keep its scan strings encoded in memory.
Some "change detection" antivirus programs add a snippet of code or data to a program in order to "protect" it. (This process is sometimes called "inoculation", but this term is also used for other antivirus techniques.) These file changes will likely be detected by other "change detection" programs, and may therefore raise a warning of a suspicious file change (see F8 for a discussion of the inadvisability of adding self-checking code to existing programs).
It is good practice to use more than one antivirus program but, by their nature, multiple antivirus programs may confuse each other!
This answer is not meant to imply that Unix viruses are impossible, or that there aren't security problems in a typical Unix environment--there are, and Fred Cohen's first experimental virus was implemented and tested on a Unix system. True viruses in the Unix environment are, however, unlikely to spread well. For more information on Unix security, see the book Practical Unix Security by Garfinkel and Spafford, O'Reilly & Associates, 1991, price $29.95 (it can be ordered via e-mail from nuts@ora.com).
There are special cases in which scanning Unix systems for non-Unix viruses does make sense. For example, a Unix system acting as a file server (e.g., PC-NFS) for PC systems is quite capable of containing PC file infecting viruses that are a danger to PC clients. Note that, in this example, the Unix system would be scanned for PC viruses, not Unix viruses. Also, any PC is vulnerable to PC MBR infectors, so special care should be taken to prevent booting a PC hosted Unix OS from a floppy infected with an MBR virus (see C12).
In addition, a file integrity checker (to detect unauthorized changes in executable files) on Unix systems is a very good idea. (One free program that can do this test, as well as other tests, is Tripwire, available by anonymous FTP from its "home" site of coast.cs.purdue.edu in /pub/COAST/Tripwire, and from several other antivirus sites.) Unauthorized file changes on Unix systems are very common, although they are not usually due to virus activity.
Note also that some machines have only 512KB or 256KB instead of 640KB of conventional memory.
One is Novell's NetWare (and possibly other network operating systems), which boots from a DOS disk and loads a "standard" DOS executable that takes complete control of the system from DOS. This executable--SERVER.EXE--could easily be infected by a DOS file infector. For example, a server's NetWare boot diskette may have to be taken from the server to a DOS PC to edit some of the configuration and startup files that have to be on that diskette. If the PC where the editing is done is infected with a file infecting virus, SERVER.EXE may well be infected when the new startup files are saved to the diskette. Such infections are virtually guaranteed to render SERVER.EXE inoperative and the server would fail at its next restart. No viruses are known to target the NetWare kernel specifically.
Another possibility is the case of a 386 (or better) system running NetWare or a self-loading OS, such as Unix, NeXTStep486, Windows NT or OS/2, since this system is still vulnerable to infection by MBR infectors (such as Stoned or Michelangelo), as these are operating system independent. Note that an infection on such a system may result in the disabling of non-DOS disk partitions (possibly beyond easy recovery) because the tricks and system conventions these viruses employ may not apply to operating systems other than DOS. The issue here is that MBR infectors are not really "DOS viruses" so much as "PC-BIOS viruses"--they can infect any machine with a PC-compatible BIOS.
Third, any OS that offers a "DOS box" or "DOS emulator" to run DOS programs can, potentially, run a virus-infected DOS program. Such activation of a virus should allow the virus to spread to any "targets" available to it under that DOS emulator. For example, a DOS program infected with a multipartite virus, when run under OS/2 would probably be able to infect other DOS executables, but not the MBR/DBS, as OS/2 only allows programs to read these critical areas of the hard drive (see E12 for more details on DOS viruses running under OS/2). With the increasing sophistication and power of computing environments, DOS emulators running on non-PC computers are increasingly available and able to run DOS viruses.
With DOS machines possibly the most common is Microsoft's SmartDrive disk cache program that came with Microsoft Windows 3.1 and subsequent versions of MS-DOS. Most versions of this software not only cache disk-reads but, by default, also cache disk-writes. This means that recently "written" files (say from saving a document in your word processor) may not have all the information about the associated file system updates written to disk by the time you exit the application, close Windows and turn off your PC. Users who simply save work then turn their PC off are even more likely to suffer from disk caching induced problems like this.
Regardless of what caused your file-system corruption, you should probably seek expert help before trying to fix anything yourself. While there are many powerful and interesting-sounding utilities of the "disk fix" kind available, all of these have the stunning ability to render your file system all but unfixable (or at least, fixable to a much lesser degree) when presented with unusual situations their authors hadn't considered when designing the programs. Unfortunately, as these programs (by definition) do not recognize these situations, they confidently pronounce that you have such-and-such a problem then ask your permission to fix it. Even when these utilities have "undo" options, they often cannot restore your file system to its originally "broken" state to give human experts their best shot at fixing it. Thus, detecting whether it is safe to let one of these programs loose on your disks is something you should normally seek expert help in deciding.
Examples: SECURE and FluShot+ (PC), and GateKeeper (Macintosh).
These programs are considered the weakest line of defense against viruses on a system that does not have memory protection, because in such an environment it is possible for a tunnelling virus (see B12) to bypass or disable them.
Examples: FindViru in Dr Solomon's AntiVirus ToolKit, Frisk Software's F-PROT, McAfee's VirusScan (all PC), Disinfectant (Macintosh).
Resident scanners: McAfee's V-Shield, and F-PROT's VIRSTOP.
Heuristic scanners: the Analyse option in F-PROT, TBAV's TbScan and ChkBoot (from Padgett Peterson's FixUtils).
Scanners are the most convenient and the most widely used kind of antivirus programs. They are a relatively weak line of defense because even the simplest virus can bypass them if it is new and unknown to the scanner. Therefore, your virus protection system should not rely on a scanner alone.
Examples: ASP Integrity Toolkit (commercial), and Integrity Master and VDS (shareware), all for the PC.
Integrity checkers are considered to be the strongest line of defense against computer viruses, because they are not virus-specific and can detect new viruses without being constantly updated. However, they should not be considered as an absolute protection--they have several drawbacks, cannot identify the particular virus that has attacked the system, and there are successful methods of attack against them too.
Examples: V-Analyst 3 (BRM Technologies, Israel), the VGUARD module of V-Care and ThunderByte's TbClean.
Note that behavior blockers and scanners are virus prevention tools, while integrity checkers are virus detection tools.
A typical PC installation might include a protection system on the hard disk's MBR to protect against viruses at load time (ideally this would be hardware or in BIOS, but software methods such as DiskSecure and Henrik Stroem's HS are pretty good). This would be followed by resident virus detectors loaded as part of the machine's startup (CONFIG.SYS or AUTOEXEC.BAT), such as FluShot+ and/or VirStop and/or ChkBoot. A scanner such as F-PROT or McAfee's VirusScan and an integrity checker, such as Integrity Master, could be put into AUTOEXEC.BAT, but this may be a problem if you have a large disk to check, or don't reboot often enough. Most importantly, new files and diskettes should be scanned as they arrive regardless of their source. If your system has DR DOS installed, you should use the PASSWORD command to write-protect all system executables and utilities. If you have Stacker or SuperStor, you can get some improved security from these compressed drives, but also a risk that those viruses stupid enough to directly write to the disk could do much more damage than normal. In this case a software write-protect system (such as provided with Disk Manager or The Norton Utilities) may help. Possibly the best solution is to put all executables on a disk of their own, with a hardware write-protect system that sounds an alarm if a write is attempted.
If you do use a resident BSI detector or a scan-while-you-copy detector, it is important to trace back any infected diskette to its source. The reason viruses survive so well is that usually you cannot do this, because the infection is found long after the infecting diskette has been forgotten due to most people's lax scanning policies.
Organizations should devise and implement a careful policy that may include a system of vetting new software brought into the building and free virus detectors for home machines of employees/students/etc who take work home with them.
Other antivirus techniques include:
Some of these systems won't prevent attack by some MBR virus infections if booting from an infected floppy. This approach is less important now, as most newer PCs allow you to change the boot order so the first hard drive is tried before any of the floppy drives.
Examples: V-Care (CSA Interprint, Israel; distributed in the US by Sela Consultants Corp.), Victor Charlie (Bangkok Security Associates, Thailand; distributed in the US by Computer Security Associates).
The workings of the first and third are probably fairly clear from these brief descriptions. The second approach works by writing special information to normally unused areas of the diskette as part of the scanning process and employing a driver in the users' machines prevents access to files that aren't marked as scanned (or to any part of a diskette that contains unscanned files). Alternatives include encrypting scanned files and drivers that only allow access to encrypted files, and so on. One advantage of this second type of system is that you only need scanners for "perimeter checking" machines, reducing the overhead and cost of keeping your scanners up to date.
Examples: D-Fence, Virus Fence, TbFence, DiskNet.
Using a layered approach, a very high level of protection/detection can be achieved with software only.
The popular idea of write-protection (see D3) may stop viruses spreading to the disk that is protected, but doesn't, in itself, prevent a virus from running.
Also, some existing hardware protection schemes can be easily bypassed, fooled, or disconnected, if the virus writer knows them well and designs a virus that is aware of the particular defense.
The big problem with hardware protection is that there are few (if any) operations that a general-purpose computer can perform that are used by viruses only. Therefore, making a hardware protection system for such a computer typically involves deciding on some (small) set of operations that are "valid but not normally performed except by viruses", and designing the system to prevent these operations. Unfortunately, this means either designing limitations into the level of protection the hardware system provides or adding limitations to the computer's functionality by installing the hardware protection system. Much can be achieved, however, by making the hardware "smarter". This is double-edged: while it provides more security, it usually means adding a program in an EPROM to control it. This allows a virus to locate the program and to call it directly after the point that allows access. It is still possible to implement this correctly though--if this program is not in the address space of the main CPU, has its own CPU and is connected directly to the hard disk and the keyboard. As an example, there is a PC-based product called ExVira which does this and seems fairly secure, but it is a whole computer on an add-on board and is quite expensive.
In some environments the Read Only attribute does provide some additional protection. For instance, under Novell Netware a user can be denied the right to modify file attributes in directories on the server. This means that a virus that infects such a user's machine will be unable to infect files in those server directories if the files have their Read Only attribute set.
Under DOS, there is no memory protection, so a virus could disable the access control system in memory, or even patch the operating system itself. On more advanced operating systems (Unix, OS/2, Windows NT) this is much harder or impossible, so there is much less risk that such protection measures could be disabled by a virus. Even so, viruses will still be able to spread, for the reasons noted above. In general, access control systems (if implemented correctly) are only able to slow down virus spread, not to eliminate viruses entirely.
Of course, it's better to have access control than not to have it at all. Just be sure to not develop a false sense of security or come to rely entirely on your access control system to protect you.
The use of the password command (e.g. PASSWORD/W:MINE *.EXE *.COM) will stop more viruses than the plain DOS attribute facility (see D5), but that isn't saying much! The combination of the password system plus a disk compression system may be more secure, because to bypass the password system a virus must access the disk directly, but under SuperStor or Stacker the physical disk will be meaningless to a virus. There may be some viruses that, rather than invisibly infecting files on compressed disks, very visibly corrupt such disks.
The main use of the "secure disk partitions" system, introduced in DR DOS 6, is to stop people from fiddling with your hard disk while you are away from the PC. The way this is implemented, however, may also help against a few viruses that look for DOS partitions on a disk.
Furthermore, DR DOS is not fully compatible with MS/PC-DOS, especially when you get down to the low-level tricks that some viruses use. For instance, some internal memory structures are "read-only" in the sense that they are constantly updated (for MS/PC-DOS compatibility) but not really used by DR DOS, so even if a sophisticated virus modifies them, it will not have any effect, or at least not that intended by the virus's author.
In general, using a less compatible system diminishes the number of existing viruses that can infect it. For instance, the introduction of hard disks made the Brain virus almost disappear; the introduction of the 80286 and DOS 4.0+ made the Yale and Ping Pong viruses next to extinct, and so on.
But remember:
However, if the LAN has lax security and is not well managed, it could help a virus to spread like wildfire. It might even be impossible to remove the infection without shutting down the entire LAN. Stories of LAN login programs, shared copies of which are run on every workstation, becoming infected are, unfortunately, not uncommon.
A network that supports login scripting is inherently more resistant to viruses than one that does not if this is used to validate the client before allowing access to the network.
Planning to minimize the impact of a virus infection on your computing is much like planning to minimize the effect of an earthquake or fire. You cannot be sure where, when or even if you will ever be "hit"; the potential impact could fall anywhere in a very wide range of possible damage; being "completely safe" can involve enormous expense; and you cannot adequately test your preparations without exposing yourself to serious risk of damage. Therefore, finalizing on the defense scheme that suits you involves deciding on the level of loss you can afford to stand and probably settling on a system that, while not "perfectly watertight," is "good enough".
Despite the importance of a good backup scheme, it is really beyond the scope of this FAQ sheet to provide a definitive guide to planning your backup procedure--that could easily take another document the size of this! All this said however, we provide the following advice as, we hope, a good starting point.
Planning an effective backup scheme really starts with answering some important questions. Consider:
When it comes to planning backup regimes with an eye to the possibility of recovering from a virus attack, you also have to consider that regularly backing-up executables (loosely, "programs") can cause problems. If you do and are infected by a virus, unless you can be absolutely sure of the date of first infection (despite sounding simple, this is not something that can commonly be done!), you may have quite a few problems finding the best backup set to restore from, as you will probably have several sets including infected executables.
For home or small business use, it may be best to maintain two kinds of backups. One would contain only your data files and one your operating system and program files (issues to consider are covered in the next two paragraphs). This may be facilitated by maintaining a strict separation of the two kinds of files, perhaps by putting the operating system and programs on one drive or partition and your data files on another. While this is probably not practical for many existing machines, enforcing adherence to the "rule" that data files should only be placed in appropriate sub-directories (folders) within a prescribed data directory may not be a bad thing.
The best way to manage backup of data files depends on the answers to too many of the questions listed above for us to give definitive advice here. While planning your backup regime, bear in mind that some viruses damage some kinds of data files, while others make small, occasional, random modifications as files are written to disk. While viruses with either of these "features" are quite rare, both of these possibilities mean that vital data files should probably be backed-up to long-cycle media sets as well as to shorter cycle sets and other steps taken to ensure you can recreate the sequence of changes. (For example, retain all transaction records so they can be re-entered.)
You should probably backup executables once after installing them and only after you are sure they are virus-free according to your current antivirus screening procedures. Never make a backup containing executables over media that hold any of your current backups. The more cautious of us maintain several cycles of executable backups. These precautions should ensure you don't face the problem outlined several paragraphs ago, and mean that should a newly installed program be infected with a virus your current defenses don't detect, you can easily restore your system and installed software to how it was before the infected software was installed, when you do become aware of its presence. You will probably have to manually reinstall any programs you installed subsequent to installing the infected program.
Having referred to this second kind of backup as "executables only", we should point out that a complete system backup is also acceptable for this type of backup. However, note that a sequence of full system backups with interim "incremental" backups (when only those files that have changed since the last complete backup are saved) is not what we are advocating. Such systems tend to be too "broad brush" to be truly useful for recovering from an unknown, future virus attack. Unfortunately, this tends to be the preferred/recommended backup scheme for small-to-medium sized systems (including most personal computers), and is typically what most popular backup software for such systems is designed to do. This doesn't mean that popular backup systems and software aren't useful, just that you have to exercise some care in using them (like excluding executable files from your incremental backups).
Having said all this, there are still a few other problems to consider, especially: Which files should you count as "data" files? This can be problematic as most people immediately think of their word-processor and spreadsheet files, and the like, as data, and that's about it. What about the files in which your programs store their configuration information? In a sense, these are as much "your data" as they are program files, because they reflect your preferred screen colors and layouts, default fonts, personalized button bars and so on. When you look at the time people spend finding the (often obscure) options settings in their programs and making them work "just right", and how upset they can be if they lose these settings, it makes sense to treat such configuration files as you treat other "personal data files" in your backup regimes. Similarly, people tend to treat system configuration files (in DOS/Windows PCs CONFIG.SYS, AUTOEXEC.BAT, WIN.INI, SYSTEM.INI at a minimum!) as part of the system, often ignoring the (sometimes considerable) fine-tuning these configuration files go through between system and executable backups.
One last point--it cannot be stressed enough that you MUST have a full, working copy of the software you need to restore your backups in a safe place. You must be able to guarantee that this software is not virus infected should you ever have to use it, AND that it is fully usable should you be facing a machine that has had its entire hard drive "wiped clean".
Most current computers will try to boot from their (first) floppy drive before trying to load an operating system off their hard disks. Because of this and the fact that every floppy disk is possibly infected with a boot sector virus, it is a very good idea to set your computer to try to boot from its hard disk. Many newer PCs offer the option to select boot order in their system CMOS setup routines. If your computer has such an option, set it to try to boot from your hard disk first.
Further, most PCs have only 64 bytes of CMOS RAM and the use of the first 48 bytes of this is predetermined by the IBM AT specification. Several BIOS'es also use many of the "extra" bytes of CMOS to hold their own, machine-specific settings. This means that anything that a virus stores in CMOS can't be very large. A virus could use some of the "surplus" CMOS RAM to hide a small part of its body (e.g. its payload, counters, etc). Any executable code stored there, however, must first be extracted to ordinary memory in order to be executed.
This issue should not be confused with whether a virus can modify the contents of a PC's CMOS RAM. Of course viruses can, as this memory is not specially protected (on normal PCs), so any program that knows how to change CMOS contents can do so. Some viruses do fiddle with the contents of CMOS RAM (mostly with ill-intent) and these have often been incorrectly reported as "infecting CMOS" or "hiding in CMOS". An example is the PC boot sector virus EXE_Bug, which changes CMOS settings to indicate that no floppy drives are present (see G8 for more details).
It might be thought that there is no point in scanning in these areas for any viruses other than those that are specifically known to inhabit them. However, there are cases when even ordinary viruses can be found in Upper Memory. Suppose that a conventional memory-resident virus infects a TSR program and this program is loaded high by the user (for instance, from AUTOEXEC.BAT). Then the virus code will also reside in Upper Memory. Therefore, an effective scanner must be able to scan this part of memory for viruses too.
Even so, note that it is not always possible to make a sharp distinction between executable and non-executable files. One person's data can be another's code and vice versa. Some files that are not directly executable contain code or data which can, under some conditions, be executed or interpreted.
Some examples from the PC world are OBJ files, libraries, device drivers, source files for any compiler or interpreter (including DOS BAT files and OS/2 CMD files), macro files for some packages like Microsoft Word and Lotus 1-2-3, and many others. Currently there are viruses that infect boot sectors, master boot records, COM files, EXE files, BAT files, OBJ files, device drivers, Microsoft Word document and template files, and C source code files, although any of the objects mentioned above theoretically can be used as an infection carrier. PostScript files can also be used to carry a virus, although no currently known virus does this.
Aside from the above, however, there is an increasing possibility of viruses spreading through the sharing of data files. More and more we see the ease with which software producers give their programs the ability to embed "objects" of many kinds into document files, and into fields in databases and spreadsheets. Perhaps the best-known of these systems are Object Linking and Embedding (OLE) in MS Windows and the OpenDoc format. As these embedded objects often have the ability to "display" themselves we see that many files traditionally thought of as data-only, will increasingly be containers carrying data and executable code. We are not aware of any virus that specifically targets such executable "objects", but it is now a trivial task to embed executable files into some kinds of document files so they will be run when the icon representing them is clicked in the finished document. There is nothing to prevent infected executables being embedded in this way, and thus for viruses to be spread through the distribution of "data files".
The more general answer is that such viruses are possible, but unlikely. Such a virus would be quite a bit larger than current viruses and might well be easier to find. Additionally, the low incidence of cross-platform sharing of software means that any such virus would be unlikely to spread--it would be a poor environment for virus growth.
A related, but different, issue is that of viruses running under operating system emulators on machines other than those for which the operating system was originally designed. This is covered in some detail elsewhere in the FAQ sheet (see C12).
Many people think that computer viruses are impossible on mainframe computers, because their operating systems provide means of protection (e.g., memory protection, access control, etc.) that cannot by bypassed by a program, unlike the operating systems of most personal computers. Unfortunately, this belief is false. As demonstrated by Fred Cohen in 1984, access controls are unable to prevent computer viruses--they can only slow down the speed with which viruses spread. If there is a transitive path of information flow from one account to another on a mainframe computer, then a virus can spread from one account to the other, without having to bypass any protections.
Consider the following example. The attacker (A) has an account on a machine and wants to attack it with a virus. In order to do this, A writes a virus and releases it. Due to the protection provided by the operating system, the virus can only infect the files writable by A. On a typical system, those would be only the files owned by A.
However, A is not alone on the system. A works with B on some joint projects. At some time, B might want to check how far A has progressed in her/his part of the project. This might involve running one of the programs that A has written--programs that are now all infected with A's virus.
On a sytem with protection based on discretionary access controls (e.g., Unix, VMS, and most other popular OSes), the program that is being executed usually runs with the privileges of the user who is executing it--not with those of the program's owner. (In the few instances where this is not the case, it presents a different kind of security threat, unrelated to viruses.) That is, when B runs A's infected program, the virus in it will run with B's privileges and will be able to infect all programs writable by B.
At some later time, A and B's boss, C, might want to check whether they have completed that joint project. Even if the boss has reasons to suspect A (e.g., as a disgruntled employee), s/he is likely to trust B and execute one of her/his programs. This results in the virus running with C's privileges (which are likely to be significantly greater than those of A and B) and infecting all programs writable by C. Quite possibly, these programs will include many owned by other employees, thus creating many more distribution chains that nobody suspects.
The virus may interfere somehow with C's normal work, which causes C (who is probably not very knowledgeable about such things as computer security and viruses) to ask the system administrator, D, for help. If D executes one of C's infected programs (and s/he is much more likely to trust a respectable person like C--who is quite probably D's boss as well--than any of C's employees), this will cause the virus that A wrote a long time ago to run with system administrator privileges and do whatever it wants with the system--infect other users' files, attack other systems, etc.
A trivial improvement of the above scenario (in terms of speeding up the virus' spread) would be for the attacker to place the virus in some kind of Trojan Horse--for example, in an attractive game or utility--placed in a publicly accessible area.
Why, then, are there so many fewer viruses for mainframe computers than for personal ones? The answer to this question is complex. First, writing a well-made mainframe virus--one that does not cause problems and is likely to remain unnoticed--is not a trivial task. It requires a lot of knowledge about the operating system. This knowledge is not commonly available and the typical youngster who is likely to hack a quick-and-dirty PC virus is unlikely to possess it or be in a position to learn it. People who possess this knowledge are likely to use it in more constructive, satisfying, and profitable ways. Second, the culture of software exchange in the mainframe world differs considerably from that of the PC world--we don't see many VMS users running around with a bootable tape of the latest game... Third, very often it is easier to attack a mainframe computer by using some security hole or a Trojan Horse, instead of by using a virus.
So, computer viruses for mainframe computers are definitely possible and several already exist (see question F1). Also, some IBM PC viruses can infect any IBM PC compatible machine, even if it runs a "real" OS like Unix. For more information, refer to questions D6 and E7.
Forms of malware other than computer viruses--notably Trojan Horses--are far quicker, more effective, and harder to detect than computer viruses. Nevertheless, on personal computers many more viruses are written than Trojan Horses. There are two reasons for this:
None of the currently available disinfecting programs do all this. For instance, because of the bugs that exist in many viruses and because some infection processes involve overwriting (part of) the objects of infection, some of the information about the original object may be irrevocably destroyed. Sometimes it is not even possible to detect that this information has been destroyed and to warn the user. Furthermore, some viruses corrupt information very slightly and in a random way (Nomenklatura, Ripper), so that it is not even possible to tell which objects have been corrupted.
Therefore, it is usually better to replace infected objects with clean backups, provided you are certain that your backups are uninfected (see D10), or from the original media. You should try to disinfect files only if they contain some valuable data that cannot be restored from backups or recompiled from their original source.
The important thing is not to avoid a certain type of software, but to be cautious of any and all newly acquired software and diskettes. Merely scanning all new software media for known viruses would be rather effective at preventing virus infections, especially when combined with some other prevention/detection strategy such as integrity management of programs.
When you perform a DIR, the contents of the boot sector of the diskette are loaded into a buffer for use in determining disk layout etc, and certain antivirus products will scan these buffers. If a boot sector virus has infected your diskette, the virus code will be contained in the buffer, which may cause some antivirus packages to produce a message like "xyz virus found in memory...". In fact, the virus is not a threat at this point since control of the CPU is never passed to the virus code residing in the buffer. Even though the virus is really not a threat at this point, this message should not be ignored. If you get a message like this, and then reboot from a clean DOS diskette (see G8) and scan your hard-drive and find no virus, then you know that the false positive was caused by an infected boot-sector loaded into a buffer, and the diskette should be disinfected before use. The use of DIR will not infect a clean system, even if the diskette it is being performed on does contain a virus (see C8 also). Please note, however, that running DIR on a diskette can result in the infection of a clean diskette if the PC is already infected.
Despite our categorical "No" answer above, there is a small risk that a virus infection could be transferred from a floppy through a DIR listing. If you use an ANSI console driver that allows key remapping, it is possible that a specially prepared diskette could reprogram your keyboard so that pressing a particular key caused an infected program on the diskette to run the next time the reprogrammed key was pressed. The risk of such an attack is very low and can easily be negated following the general advice for preventing ANSI bombs (see B14).
Mac users with system software prior to version 7.0 should be aware of a greater threat in their environment. Various system resources (which can contain executable code) are loaded from the automatic access to a diskette that is part of the system building its desktop view of the diskette's contents. When such a resource is required, the most recently loaded one will be used. Thus, if a diskette with a virus-infected resource in the Desktop file is in your Mac's drive, and an uninfected copy of that resource has not subsequently loaded from elsewhere, the next time that resource is required the infected copy will be executed, along with the virus. This kind of attack was removed with the introduction of version 7.0 (and later) of the system software, which handles such things quite differently. A common Mac virus, WDEF, uses this infection path, as do a few others.
Early versions of AmigaDOS are susceptible to a threat similar to the Mac WDEF virus--on inserting a diskette into the drive, the operating system runs the Disk Validator from the diskette. At least one Amiga virus, Saddam, attaches itself to Disk Validator to help it spread. Version 2.0 of AmigaDOS eliminated the threat of this type of attack by removing the need for the Disk Validator.
Also, bear in mind that generally all DOS sessions share the same copy of the command interpreter. Hence if it becomes infected, the virus will be active in all DOS sessions.
Virus-resistant however, is by no means virus-proof. For instance, most of the well-behaved resident viruses that infect only COM files (Cascade is an excellent example), will work perfectly in a "DOS box". All non-resident COM infectors will be able to run and infect too. Aside from DOS viruses, MS Windows users can also contract several currently known Windows-specific viruses, which are able to infect Windows applications properly (i.e., they are compatible with the NewEXE file format).
Any low level trapping of Interrupt 13, as by resident boot sector and MBR viruses, can also affect Windows operation, particularly if protected disk access (32BitDiskAccess=ON in SYSTEM.INI) is used.
There are a few provisos to be made. If your computer uses ANSI screen and keyboard controls, you may be susceptible to an ANSI bomb (see B14). An ANSI bomb may, merely by being placed in text read on the screen, temporarily redefine keys on the keyboard to perform various functions. It is, however, very unlikely that you will ever see an ANSI bomb in e-mail, or that it could do significant damage while you are reading mail.
Another possibility is that mail can be used to send programs. To do this program files have to be encoded into a special form so the binary (8-bit) program files are not corrupted by transfer over the text-only (7-bit) e-mail transport medium. Probably the commonest of these encoding schemes is uuencoding, though there are several others. If you receive an encoded program, you normally have to use a decoding program or special option in your e-mail program to extract it and decode it before it can be run. Once you have extracted the program though, you should then treat it as you would any other program whose source you do not know, and test it before you run it.
A third possibility is with the newer, highly-automated online systems. Some of these attempt to make online access much easier for the user by automating such features as file transfer and program updates. At least one commercial online service is known to have the capability of sending new programs to the user and to invoke those programs while the user is still online. While there is no reason to assume that any service that does this will infect you, any time things are going on that you are not being told about, you are at greater risk.
GIF and JPEG (.JPG) files contain compressed graphical information. Every now and then, rumors arise that is possible to infect those files with a virus in such a way, that it will spread when you display one of these images. This is technically impossible--no part of the GIF or JPEG format contains code that is executed by the viewer program.
It is possible to use the least significant bit of the color information for each pixel in GIF files to store additional information, without visibly altering the quality of the picture contained in the file. This is called "steganography" and is sometimes used to transmit secretly encrypted messages. Since a virus is nothing more than information, it is possible to "encode" it into a GIF file and transmit it this way. However, the recipients must be aware that the GIF file contains such hidden information and take some deliberate steps to extract it--it cannot happen against their will.
Further, some antivirus researchers have samples in their collections that they count as viruses, but that several other experts strongly deny are viruses. Sometimes these are "partial viruses", where a virus has not properly infected a host and are therefore non-infective, other times they are well-known non-viruses. As some of these non-viruses are known to be in some of the common test sets, some antivirus software vendors count them amongst the viruses they detect.
As of January 1995 there were about 5,600 PC viruses, about 150 Amiga viruses, about 100 Acorn Archimedes viruses, about 45 Macintosh viruses, several Atari ST viruses, a few Apple II viruses, four Unix viruses, three MS Windows viruses, at least two OS/2 viruses and two VMS DCL-based viruses.
Fortunately, few of the existing viruses are widespread. For instance, only about three dozen of the known PC viruses cause most of the reported infections and fewer than 200 PC viruses have been found in the wild at all.
Those that do spread widely are able to do so for a variety of reasons. A large target population--millions of compatible computers--helps. A large virus population helps. Vendors whose quality assurance relies on, for example, outdated scanners, help. Users who gratuitously install new software on their systems without making any attempt to test for viruses help. All of these things are factors.
There are two things to watch out for here: the general "style" of the software, and the scan strings that scanners use to identify viruses. Scanners should be updated more frequently than other software, and it is probably a good idea to update a scanner's set of scan strings at least once every two months. In the six or so months prior to January 1995, most of the popular PC-based virus scanners typically added detection of about 500-600 new viruses or variants--this averages out to between two and three new viruses per day!
Some antivirus software looks for changes to programs or specific types of viral "activity", and these programs generally claim to be good for "all current and future viral programs". However, even these programs cannot guarantee to protect against all future viruses, as new "attack" and anti-antivirus methods are continually being developed by virus writers. Thus, even this type of antivirus software needs to be upgraded occasionally.
Of course, not every antivirus product is effective against all viruses, even if upgraded regularly. Thus, do not depend on the fact that you have upgraded your product recently as a guarantee that your system is free of viruses!
There are three ways in which virus simulators are usually used:
1) For educational purposes. The second kind of virus simulators are very useful and valuable for this purpose, provided the simulated environment is realistic enough. The first kind are also somewhat useful--mainly teaching the users what the video- or audio-effects of particular viruses are like. There is the danger, however, that users will get the incorrect impression that every computer virus demonstrates itself in some visible or audible way. The third kind of virus simulators are not useful for this purpose--they do not show how computer viruses work, do not show what computer viruses do, and because their virus fragments are not reliably detected as viruses by many good scanners, may give the wrong impression of a scanner's value.
2) As an installation check that antivirus defenses are installed and working. The first and second kinds of virus simulators are unsuitable for this, because they do not trigger any antivirus defenses. Even the third kind of virus simulators have a rather limited value in this regard, as the files generated by them often fail to trigger virus defenses, which are designed to protect against real viruses. Unlike the producers of such simulators, many believe it is the job of the producer of an antivirus product to provide the means of checking whether their product is installed and working. This position is based on the authors knowing their products better than anyone else and that updated check methods will normally be provided as the antivirus defenses employed in any given product change.
3) As a test of the quality of the antivirus defense--usually a scanner. Again, the first two kinds of simulators are unsuitable for this purpose because they do not trigger antivirus defenses. The third kind of virus simulators often do, from which many users get the impression that they are suitable for these testing purposes. This is a serious misconception. The files that such programs generate are not real viruses; antivirus programs, particularly virus-specific ones like scanners, are designed to detect real viruses. Therefore, one must not draw a conclusion from the ability or the inability of a product to detect "simulated viruses" of the third kind--the fact that they are detected does not necessarily mean that a real virus will be detected, and the fact that they are not detected does not mean that the real virus it is supposed to represent will not be detected!
One exception to the above are simulators that do not generate files containing scan strings, but which simulate the different kinds of attacks that real viruses use, but without being able to replicate. Examples of such attacks include different methods of tunnelling, stealth, attacks against integrity checkers, and so on. Such simulators are useful for testing antivirus products that are not virus-specific, especially if the simulator exercises a wide range of known attacks.
By definition (see B1), viruses do not have to do something "bad" (although many people argue that the uninvited "resource wasting" that is almost inherent in viral activity is necessarily bad). From this point (and based on his somewhat esoteric definition of the term computer virus) Fred Cohen has argued that "good" or "useful" computer viruses are a serious possibility. In fact, Dr. Cohen offered a reward of $1000 for the first clearly "useful" virus--despite several potential claimants, however, he hasn't paid up.
Although there has never been a position that was widely agreed upon as a result of any of these discussions, many contributors to this forum believe that there are serious problems with the idea of implementing useful computing functionality through self-replicating programs. Vesselin Bontchev's paper originally delivered at the 1994 EICAR conference, titled "Are `Good' Computer Viruses Still a Bad Idea?", is available by anonymous FTP from ftp.informatik.uni-hamburg.de (IP = 134.100.4.42), as pub/virus/texts/viruses/goodvir.zip. Anyone wishing to raise this discussion in Virus-L/comp.virus again should read and carefully consider this paper before posting. It contains many strong arguments against the idea of "good computer viruses", and some prescriptions of how good viruses would have to be implemented and distributed to deserve the label "good". To date no strong arguments countering the points in this paper or otherwise arguing in favor of the concept of good viruses have been posted to the group.
A simple and intuitively attractive idea--in fact, some antivirus programs have included options to do just this. There are, however, some serious flaws with this approach.
This method cannot prevent the program from getting infected in the first place. Further, if a program that has been protected this way becomes infected later, whenever it is run the virus code will be activated first. The virus may then be able to detect or even remove the self-checking code, or it might make it totally ineffective by using stealth techniques, so the self-checking code only "sees" the original, non-infected program.
Some programs contain an internal self-check--much antivirus software, for example. Such internal code might also be unable to detect stealth viruses, but unless the external self-check code uses stealth techniques too, the result will be a conflict, where the internal check will notice the newly added code and decide that it has been "infected".
Moreover, this method is ineffective against "companion" viruses that don't modify the applications they infect.
It may not be possible to protect all programs this way. For example, under DOS it is relatively easy to add code of this type to most COM files (unless the original program was slightly less than 64K, and the resulting file would break that limit). However, EXE files are more of a problem--especially those containing internal overlays, where one cannot append the code to the file, as the resulting file might become too big to load. Windows applications are also a problem, as they have two different entry points, and special care has to be taken to handle that correctly.
On the other hand, adding internal self-checking to programs as part of their development is a good idea. Although it has the same limitations regarding stealth viruses, it does not cause the conflicts described above, and can be put in any program at compile-time. It is also much more difficult for viruses to bypass.
It is possible, if you have an urgent need to check a system when you don't have any antivirus tools, to run CHKDSK or MEM and note down the values reported (see C1) and then to boot from a known clean system diskette and compare the results returned by CHKDSK or MEM. If the total amount of conventional memory reported is different between the two boots then you may have a viral problem but this information alone cannot tell us if it is Stoned. If you cannot see the PC's hard disk (usually the C: drive) then it is even more likely you have a virus problem, though definitely not Stoned. If you have a "disk editor" type program, looking at the boot sector of a suspect floppy, or the MBR of the suspect hard drive may be helpful. If you have Stoned, the first byte will indicate the characteristic far jump of the virus (hex: EA) instead of the more common short jump (hex: EB) of the boot loader. Even if that is the first byte, you could be looking at a perfectly good disk that has been "inoculated" against the virus or is infected with some other virus which makes similar changes, or at a diskette that seems safe but contains a totally different type of virus.
Further, versions of Stoned with no message whatsoever or only the leading bell character have become very common. These versions of Stoned are likely to go unnoticed by all but the most observant, even when regularly booting from infected diskettes.
Contrary to some reports, the Stoned virus does not display the message "LEGALISE MARIJUANA", although such a string is quite clearly visible in the boot sectors of diskettes and MBR's of hard disks infected with the "original" version of Stoned.
The first of these viruses that infects the computer will overwrite the Master Boot Record with its body and store the original MBR at a certain place on the disk. So far, this is normal for a boot-record virus. But if now the other virus infects the computer too, it will replace the MBR (which now contains the virus that has come first) with its own body, and store what it believes is the original MBR (but in fact is the body of the first virus) at the same place on the hard disk, thus overwriting the original MBR. When this happens, the contents of the original MBR are lost. Therefore the disk becomes non-bootable.
When a virus removal program inspects such a hard disk, it will see the second virus in the MBR and will try to remove it by overwriting it with the contents of the sector where this virus normally stores the original MBR. However, now this sector contains the body of the first virus. Therefore, the virus removal program will install the first virus in trying to remove the second. In all probability it will not wipe out the sector where the (infected) MBR has been stored.
When the program is run again, it will find the first virus in the MBR. By trying to remove it, the program will get the contents of the sector where this virus normally stores the original MBR, and will move it over the current (infected) MBR. Unfortunately, this sector still contains the body of the first virus. Therefore, the body of this virus will be re-installed over the MBR ad infinitum.
There is no easy solution to this problem, since the contents of the original MBR are lost. The only solution for the antivirus program is to detect that there is a problem, and to overwrite the contents of the MBR with a valid MBR program, which the antivirus program has to provide itself. If your favorite antivirus program is not that smart, consider replacing it with a better one, or try using the boot sector disinfection procedure described elsewhere (see C3).
In general, infection of the same file or area by multiple viruses is possible and vital areas of the original may be lost. This can make it difficult or impossible for virus disinfection tools to be effective, and replacement of the lost file/area will be necessary.
Part of Flip's infection routine decrements by six the "total number of sectors" field in the BIOS Parameter Block (BPB--a table of critical disk geometry data) in the DOS boot sector of the boot partition. For partitions of 32MB and under this field is meaningful, but in larger partitions, this field is set to zero and a field in the "extended BPB" contains the "big number of sectors" for that partition instead. Not knowing about larger partitions, Flip renders the large partitions it meets a shade under 32MB. The fix for this is to use a disk sector editor to set the word at offset 13h of the affected DOS boot sector to "00 00" (they should be set to "FA FF" if the situation above applies). If you don't understand these instructions, do not attempt to follow them and seek the help of a more technically knowledgeable person.
The facetious answer aside, the real question here is usually more one of "How do I ensure I have a clean boot floppy?"
As with so many issues concerning viruses, the important thing is to be prepared in advance. As with backups, a current, clean boot disk should be a standard part of every personal computer system, as there are other occasions than when facing a real or suspected virus infection where being able to boot your computer to a "known good" state are useful or desirable (e.g. you accidentally delete your disk-compression driver from your hard disk). As with backups, a current, clean boot disk is one of the standard parts of a personal computer system most commonly missing.
The important thing in preparing a clean boot diskette, especially where it has to be used with a (suspected) virus infection, is that it must not run a single byte of code from your hard disk. This means your boot floppy must contain all the basic operating system files, device drivers and configuration commands necessary to make your system minimally usable. This diskette must be prepared on a system that is, itself, guaranteed "clean" and it should be write-protected immediately after it is completed. Aside from a basic, minimal operating system, your emergency boot diskette should contain the utilities necessary to install your OS to a hard disk and basic diagnostic or "fix it" programs and your favorite antivirus tools. Depending upon disk space considerations, you may need additional diskettes to hold all these utilities. For example, if you use DOS it is a good idea to copy the following utility programs to your emergency boot disk (if your version of DOS includes them): FDISK, CHKDSK and/or SCANDISK, FORMAT, SYS, MEM, UNFORMAT, UNDELETE, MSD.
When it comes to rebooting your computer from a clean system disk, it is most important that you perform a "cold start". On a PC, this means pressing the reset button or turning the power off on again, not by pressing Ctrl-Alt-Del. Regardless of the machine type, if you are unsure, use the power off then power on method just described. It is even more important that your machine is correctly configured to try booting from the floppy first. Most contemporary BIOSes have an option to select the boot order (A: then C: or C: then A:)--this must be set to A: then C: for this procedure, though normally we strongly recommend that you set this option to C: then A:.
As systems change from time to time, you may occasionally need to update this most critical of diskettes so it will still boot your system to a usable state. As you may have recently contracted a new virus that bypasses your current antivirus precautions, this update process can put you at risk of infecting your "clean" emergency boot diskette. Because of this, it is prudent to have two such diskettes. With system changes you would update these in a "leap frog" manner. This means your previous emergency boot diskette might still bring your machine up to a minimally useful state (such that you may still be able to make repairs) should your updated emergency boot diskette be infected by a previously unknown virus.
Unfortunately, this isn't the whole story either! A PC virus known as EXE_Bug can fake out the boot process by setting the PC's CMOS to look as if there are no floppy drives in the machine. Most BIOS'es don't even try to boot from a floppy in this case, and go straight to the hard disk, loading the virus from the MBR. When EXE_Bug first loads into memory, it checks to see if there is a diskette in the first floppy drive, and if there is, it loads the boot sector from the diskette and lets the floppy boot as normal. Most people don't notice the subtly different boot time and drive access order involved in this, so they think they have booted clean, when in fact the virus is active in memory! To circumvent this possibility, you have to check the PC's CMOS settings before letting the floppy boot proceed, make sure that your PC "knows" it has a floppy drive, and, with some PCs, make sure that the boot order option is set to "A: then C:". This presents a chicken-and-egg situation on some machines, as you may have to boot DOS on the machine to be able to run the utility program that lets you change its CMOS settings.
Remember, if you changed your BIOS's boot order option, set it back to C: then A: after disinfecting your PC.
Generally, you would be better to use "DIR /P" instead of "DIR | MORE", as this avoids the creation of the temporary files. If you use an alternative commandline interpreter, none of the above may apply.
Unfortunately, one of the earliest and most widely available implementations of this idea prints a message on screen at each system startup to the effect "ChipAwayVirus installed". This is supposed to calm the owner's nerves, making them confident that their BIOS antivirus system is working for them. For fairly obvious reasons, it tends to have the opposite effect!
[End of Virus-L/comp.virus FAQ sheet]