Allow F-Secure Anti-Virus to disinfect the relevant files.
For more general information on disinfection, please see Removal Instructions.
Eliminating a Local Network Outbreak
If the infection is in a local network, please follow the instructions on this webpage:
Swen is a worm that replicates via email, local network (LAN), IRC and Kazaa. It uses a vulnerability in Internet Explorer to execute directly from e-mail. Swen worm appeared on 18th of September 2003. It is most likely written by the author of Gibe worm (Begbie) and this worm has similar features as the latest Gibe variants.
The worm's file is a Windows PE executable 106496 bytes long. It is not compressed by any file compressor.
When the worm's file is run, it checks whether it's already installed and if not, it copies its file to Windows directory with a random name (for example MLMHP.EXE) and creates a startup key for this file in the Registry:
where is the name of the worm's file. This way the worm's file is always started with Windows.
If the worm is already installed on a computer, it shows the following messagebox:
Otherwise the worm shows the following messagebox:
If a user clicks 'No' button, the worm installs itself to system hiddenly. If a user clicks 'Yes' button, the worm shows a fake installation dialog:
and after some time it reports successful installation:
During installation the worm creates a batch file that has a name of an infected workstation. This batch file contains the following text:
where [name] is the name of the worm's executable file.
The worm extracts the list of SMTP and NNTP servers from its body into the SWEN1.DAT file that is placed into Windows directory.
Then the worm modifies default startup keys for BAT, SCR, EXE, REG and PIF files in the Registry:
As a result, the worm gets control every time a user tries to run executable and registry files.
Additionally the worm disables Registry tools by creating the following key:
As a result a user will not be able to run Regedit utility and import REG files data. The worm will show the following messagebox in such case:
The numbers in this messagebox are randomly-generated.
The worm creates a set of subkeys in the following key:
These subkeys contain information about SMTP server, user's e-mail, key name of installed worm's file, name of infected computer user, name of a zip archive that the worm tries to create using WinZip, name of mIRC folder and some other data.
During installation process the worm enables sharing for Kazaa client, copies itself several times into Kazaa shared folders and also replaces SCRIPT.INI file of mIRC client with the one that sends out the worm's file to every user joining a channel where an infected user is present. The worm also copies its file to startup folders of remote computers via network.
Propagation (local network)
The worm attempts to spread itself via local network (LAN). It looks for mapped network drives, accesses them and if it finds the following directories in the root folder:
it copies its file with a random name to the following folders:
As a result remote computers will become infected with the worm after they are restarted.
The worm creates its own SCRIPT.INI file in mIRC installation folder. This script makes an IRC client send a file called 'WinZip installer.zip' to every user joining a channel where an infected user is present.
The worm modifies the Registry to enable sharing for Kazaa client, then it locates Kazaa shared folder and copies itself there with a generated name. The name is generated from the following strings:
These files can have EXE or ZIP extensions.
Propagation (E>-mails and newsgroups)
The worm periodically scans HTML and ASP files on a hard drive and stores found e-mail addresses in the GERMS0.DBV file located in Windows folder. The worm also reads .EML, .DBX, .WAB, and .MBX files and fetches e-mail addresses from there. The worm does not fetch addresses containing 'delete' and 'spam' strings.
The worm also can search for e-mail addresses in various newsgroups. It connects to NNTP servers listed in the SWEN1.DAT file, gets a list of all newsgroups on that server and searches recent messages in these newsgroups for 'nfrom:' and 'nreply-to:' tags. When such tags are found, the worm gets e-mail addressed after them and writes them to the GERMS0.DBV file. This way the worm can harvers a lot of e-mail addresses to send itself to.
The worm can post its e-mails to newsgroups, the names of which it finds during searching process. The worm sends the same kind of messages as it sends via e-mail.
The worm reads SMTP server address and user name from the Registry. However, if it can't find this info, it shows a fake MAPI error dialog asking a user to input that data:
The worm sends itself a very legitimately-looking messages that are composed from different text strings hardcoded in the worm's body. It also checks the current date and uses the current month inside the text of the email message. On that way it will spread with different messages each month of the year.
Here is an example of such message sent in September:
The attachment name, subject and part of the infected message is randomly composed from text strings hardcoded in the worm's body.
The fake sender's address is selected from the following parts:
The domain name for these e-mails is selected from the following parts:
The domain suffix for these e-mails is selected from the following parts:
followed by one of the following:
The fake recipient's address is also composed from the above shown strings, however the fake recipient's name is selected from the following parts:
The subject is composed from the following parts:
The worm is usually attached to infected messages as an EXE file. The attachment name is randomly generated from numbers and the following parts:
For example the infected attachment name can be Q591362.EXE or UPDATE98.EXE. The IFrame exploit is not present in such messages. In some cases the worm's attachment can be in a ZIP archive.
The worm can also compose fake forwarded or bounced e-mails from the following parts:
The bodies of bounced e-mails can have the following text strings:
Such e-mails usually contain IFrame exploit and the worm's file with PIF, BAT, COM, SCR or EXE extension and there is no Microsoft-like looking message body in them. The IFrame exploit allows the worm's attachment start automatically on older or unpatched versions of certain e-mail browsers.
The worm terminates processes of security and anti-virus software that have the following strings in their names:
The worm also doesn't allow to start files that have the above strings in their names. When such file is being started, the worm shows the following messagebox and stops execution if such file:
The numbers in this messagebox are randomly-generated.
If the worm finds a debugger in a system, it shows a messagebox with the following text:
The worm keeps its own counter on a certain webpage. Every infected computer tries to access that page and that increases the counter there. By the time of this description creation (18th of September 20:00 GMT) the counter value was over 510000, but we believe that this is not the actual number of infected computers.
This minor variant was found on 9th of October, 2003. It has been created by compressing the original virus with UPX. This has shrunk the virus from 106496 bytes to 52224 bytes, making it undetectable to some antivirus programs.
In addition, many references to Microsoft in the original virus have been changed to references to Tiscali, an Italian ISP.
F-Secure Anti-Virus detected this modified version of the virus without any need for updates.
This minor variant was also found on 9th of October, 2003. Like the previous variant this one is also compressed with UPX file compressor. The packed file size is 52224.
Swen.C has a bit different set of text strings mentioning both Tiscali and Microsoft and also the name of Tiscali's CEO Renato Soru. A few Tiscali links that were present in the B variant were slightly modified.