Issue No.10 - October (2004 vol.37)
Published by the IEEE Computer Society
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MC.2004.173
New Chips Stop Buffer Overflow Attacks
Chip makers are designing a new generation of microprocessors to stop buffer overflow assaults, exploits that hackers often use to attack and extract data from PCs or servers.
AMD's Athlon-64 chips for notebook and desktop computers and its Opteron processors for servers include features that provide buffer-overflow protection. However, the feature won't work until it is supported by operating systems, said Richard Brunner, an AMD Fellow in software architecture. "We've been working with [OS] vendors to get them to take advantage of the hardware features," he explained.
Intel offers buffer-overflow protection in its Itanium chips for servers and will do so in future chips. Microsoft will support the buffer-overflow features in its Windows Server 2003 service packs scheduled to be released in the first half of 2005. Different Linux flavors also plan to work with this technology.
A buffer overflow occurs when a program or process tries to store more data in a buffer than it was designed to hold. The extra information can overflow into other memory sections, corrupting or overwriting data stored there. Systems could then execute malicious instructions inserted by a hacker into the data. The Slammer and Blaster worm PC attacks in 2003 and the Slapper worm that infected many Linux-based Web servers in 2002 involved buffer overflows.
Operating systems that support the buffer-overflow prevention approach mark certain data in memory with a bit that identifies them as executable or nonexecutable. If hackers try to force a computer to execute code from a section of memory designated as nonexecutable, the processor will send an OS exception that causes Windows to generate a page fault that closes the application.
The AMD chips let users turn off the new security feature for legacy programs not written to work with the technology so that the applications can continue to function. Intel chips will work in a similar manner, according to company spokesperson Howard High.
This type of protection is important, and OS vendors should support it, noted Carl D. Howe, cofounder and analyst at Blackfriars Communications, a market research and consulting firm.
According to Howe, buffer-overflow prevention currently is limited almost exclusively to Intel-compatible systems, although there are a few other implementations.
News Briefs written by Linda Dailey Paulson, a freelance technology writer based in Ventura, California. Contact her at firstname.lastname@example.org.
Two Companies Make the Hard Drive Wireless
A company has developed a device that utilizes wireless universal serial bus (USB) interconnect technology to let multiple users write to and access data on the same external hard drive. This can help users in small companies or households share data without setting up expensive storage area networks or installing cabling.
The Network Storage Link by Linksys, a Cisco Systems division, connects to an external hard drive on one end and to a router on the other. Multiple PCs or laptops with wireless adapters or cards could then connect to files stored on the drive via the router and the NSL.
A machine without a wireless adapter or card could use a USB cable to connect to the NSL and thus communicate with an external drive, noted Linksys spokesperson Karen Sohl. Or a user could connect a computer without wireless capabilities to a router via an Ethernet cable. The user could then communicate with an external drive via the router and the NSL, which contains USB and Ethernet ports.
For cable-free data communications, the NSL uses wireless USB, which works with the same protocol and architecture as a traditional wired USB except that it runs over ultrawideband. UWB is a low-power, short-distance, radio-based technology that transmits large amounts of data over a wide range of frequency bands.
Linksys has teamed with drive maker Maxtor to develop a common set of instructions so that their products will work together. However, the NSL will work with any vendor's external drive that also uses the instruction set, said Sohl.
NSL includes an optional file management system that coordinates enhanced tasks such as setting up user permissions and giving users their own sets of files to work with.
Philip Marshall, director of wireless-technology research for the Yankee Group, a market analysis firm, said it is not clear yet how big a market there will be for the new technology.
Spam-plagued Researchers Develop Antispam Technique
Computational-biology researchers have used a pattern-discovery application created for protein studies to develop an algorithm that enables e-mail systems to automatically and accurately identify spam.
Isidore Rigoutsos and Tien Huynh of the Bioinformatics and Pattern Discovery Research Group at IBM's T.J. Watson Research Center developed the Chung-Kwei system, named after a mythical Chinese figure known for protecting valuable property from evil.
During testing, the system's main algorithm—which is part of IBM's larger SpamGuru project ( www.research.ibm.com/spam/filtering.html) —correctly identified 96.5 percent of incoming spam messages and falsely identified only one out of 6,000 non-spam messages.
The two IBM researchers developed Chung-Kwei based on the Teiresias pattern-discovery algorithm that Rigoutsos created in 1996 to look for patterns in protein and nucleic acid sequences and in other strings of biological material. Chung-Kwei analyzes the patterns of characters used in both spam and regular e-mail and generates a database of patterns found only in spam.
The database can grow as the system analyzes more e-mail. The system examines e-mail for patterns also found in its spam database. It then computes a score for messages based on the number and prevalence of spam-related character patterns they contain. The system identifies messages as spam based on their scores.
Individual users or network administrators can train the algorithm by exposing it to new spam messages with different character patterns so that it won't be fooled by spammers' tricks, such as replacing the "S" in words commonly used in spam with an "$."
To improve Chung-Kwei's filtering performance, researchers trained it on 66,000 spam and 22,000 nonspam messages, Rigoutsos said. In this process, researchers used some of the many unsolicited messages they regularly receive.
Current antispam applications generally use several techniques to recognize unsolicited mail, such as identifying whether the sender's address has been spoofed or finding a sender's presence on lists of known spammers or trusted contacts.
Paul Hoffman—director of the Internet Mail Consortium, which hosts many mailing lists related to Internet mail standards—expressed no confidence in spam filtering. "We haven't seen a filtering technology that spammers haven't figured out how to circumvent," he said.
But John R. Levine, chair of the Internet Research Task Force's Anti-Spam Research Group, said the Chung-Kwei approach is "a very clever way to do mail filtering within the limits of what mail filtering can do. But spam filtering at its best is only a stopgap. Even really good filtering makes mistakes now and then."
IBM says it plans to commercialize the system eventually, probably in its Lotus software products. First, however, Rigoutsos said, the company must test it in a live, real-world setting.