, SecurityFocus 2005-09-13
Microsoft's decision to cancel a security fix after finding problems with the patch has security experts questioning whether waiting for the fix to come next month might leave them open to attack.
The concerns come after Microsoft announced last Thursday that a critical fix for the Windows operating system would be distributed in the following week. The next day, the software giant pulled the planned patch due to quality issues, according to Mike Reavey, operations manager for the Microsoft Security Research Center.
"Late in the testing process, we encountered a quality issue that we decided was significant enough that it required some more testing and development before releasing it," Reavey said in a posting to the MSRC Blog. "We have made a commitment to only release high-quality updates that fix the issues at hand, and therefore we felt it was in the best interest of our customers to not release this update until it undergoes further testing."
The few details that the software giant has provided--the flaw is a critical bug in Windows and does not require a reboot to fix--will not likely help would-be flaw finders to narrow their search. If the company had actually released a flawed patch, attackers could have reverse engineered the fix to find the original flaw. Since no real details of the issue were published, however, there is little danger, a spokesperson for the software giant said.
Yet, the move has left network administrators feeling vulnerable. The knowledge that a critical flaw is being left untended has security researchers second guessing whether Microsoft plans to release the patch next month, and if so, has the company's focus on regularly scheduled patching put them in danger.
"There's knowledge of a flaw and, because (Microsoft) can't meet the deadline of the next few days, they're going to delay it a month," said one member of the DShield mailing list. "So from a security point of view, we have a hole that is known but not patched."
The person who posted the criticism did not immediately respond to requests for further comment.
Microsoft has not specified when the company plans to release the patch. The release schedule will be determined by "customer need," a spokesperson said on Tuesday.
While another month's ferment may not make the current vulnerability more threatening, the move towards scheduled patches generally makes corporate customers less secure, said Marc Maiffret, chief hacking officer for eEye Digital Security and a critic of scheduled security updates.
"The monthly schedule doesn't make the customers more secure," he said. "Microsoft is doing it more for customer convenience than for customer security."
Maiffret also said that fewer announcements of vulnerabilities means that Microsoft's operating system security is under the microscope less often, resulting in less pressure to get flaws fixed.
"Almost every other major software company is still able to produce a patch in a short time, but Microsoft takes six months or more," Maiffret said.
eEye keeps a running total of how long companies have known about certain vulnerabilities. Microsoft has mulled the vulnerability at the top of the company's list for almost six months, according to eEye's Web site.
A Microsoft representative was not available for an interview. However, in his posting to the MSRC blog, Microsoft's Reavey said that large-scale testing meant that patches could sometimes be delayed.
"When we moved to a monthly release cycle almost two years ago, we planned for a significant focus on testing," Reavey stated. "That focus means that sometimes the testing process and our decision to only release quality updates might mean a month without any updates."
Other members of the security community lauded the regular schedule introduced by Microsoft, arguing that giving due notice means that patches are more likely to be applied and that makes for better security.
"In my scheduled time with limited resources, I allocate a certain amount of time to patching systems," said another network administrator that posted to the DShield security mailing list. "I may not want to do an out-of-band or ad-hoc deployment of a critical patch that is not related to a virus outbreak or worm. I understand the day may arise where 0-day worms are created. However, until such time I am going to stick to my schedule."
The person who posted the comment did not immediately respond to requests for an interview.
Microsoft is not the only company to move to regularly scheduled security updates. Database maker Oracle has also made the move and has also, coincidentally, become the target of criticism for taking too long to produce fixes. In July, security researchers claimed that the company took almost two years to produce a security fix. An Oracle representative was not available for comment.
Such problems can hurt a company's bottom line, according to recent research that has statistically shown many software makers suffer a decline in stock price when vulnerabilities are announced.
It's a debate that is unlikely to go away, said Bruce Schneier, chief technology officer for network protection firm Counterpane Internet Security and a well-known security expert.
"This is the Catch-22 for software vendors," Schneier said. "A badly written, badly tested patch would be worse than the attack. Microsoft has to get it right. The problem is that they also have to get it fast."