Microsoft has admitted knowing about a still-unpatched Internet Explorer (IE) bug for more than a year, but defended its security process against critics.
According to Mike Reavey, director of Microsoft's Security Response Center (MSRC), the company first got word of a critical flaw in an ActiveX control in early spring 2008. The bug can be exploited through IE6 and IE7 on Windows XP. Two researchers, Ryan Smith and Alex Wheeler, reported the bug to Microsoft when they worked together at IBM's ISS X-Force in 2007. Smith is now a vulnerability researcher at VeriSign iDefense, while Wheeler manages 3Com's TippingPoint DVLabs.
Although both Smith and Wheeler have declined to say when they reported to vulnerability, the bug's CVE (Common Vulnerabilities and Exposures) number pointed to an early 2008 reporting date.
The 16- to 18-month stretch between early 2008 and now is too long for Microsoft's customers to go without a patch, said John Pescatore, Gartner's primary security analyst. "That's just not an acceptable timeframe," Pescatore said. "It shouldn't take a year, not [for] a company the size of Microsoft.
"It's really hard to think of some technical reason why it would take 18 months. That means it must be for other reasons, business reasons or product reasons or priority reasons," he said. "But this had to have been pretty high-priority."
"We kicked off our investigation as soon as the vulnerability was reported to us," countered Reavey. "When a vulnerability is reported, we not only look at that, but also investigate other issues around it to provide as much protection as possible."
The 16- to 18-month time span, however, is certainly above average, Reavey agreed. "The timeline is not the norm," he said. "The vast majority of vulnerabilities are patched before there's ever an attack."
What, then, took so long?
Although Reavey declined to get specific today, Smith, one of the researchers who reported the vulnerability, hinted at reasons. "The nature of this flaw is sort of unique," he said. "The mechanics of this are sort of unique as well. It was those unique qualities that required more time than Microsoft would normally need."
Smith refused to criticise Microsoft for not patching sooner. "All along the way, they've told me how far things have progressed," he said of Microsoft's security team. "They would ping me every time they reached a milestone on the fix."
Even so, he admitted that patching quickly is better than fixing slowly. "As a security researcher, you always want to see a patch the day after you report a bug," Smith said.
In fact, Microsoft has not yet wrapped up work on a fix, Reavey acknowledged. "We'll release something that will block all known attacks next week," he said, referring to 14 July, the next Patch Tuesday, when Microsoft rolls out its monthly security updates. But it won't be a full-fledged patch.
Instead, next week's updates will set 45 "kill bits" in the Windows registry, disabling the ActiveX control. Earlier this week, Microsoft published a free tool that did the same thing, but the tool required someone to sit at each PC, browse to a support site, download the tool and then run it. "That just wasn't realistic for enterprises," said Gartner's Pescatore. "It was 'high touch,' and certainly not something that, say, Procter and Gamble could do."
Microsoft did consider issuing the "kill bit" update earlier as a protective measure, both Smith and Reavey said. "They did, but they wanted to deliver the best patch," said Smith.
Reavey gave essentially the same reason why Microsoft didn't take action earlier. "We always want to give customers a complete solution," Reavey said, alluding to a patch rather than the automated workaround it will issue next week. "If we had tried to do something earlier, it wouldn't have been as clean for customers."
He also denied that Microsoft had known that attacks were out and about last month, as others have claimed. IBM's X-Force, where Smith and Wheeler worked when they discovered the vulnerability, claimed that attacks had been recorded as early as 11 June.
"We were made aware of the attacks the day before we released the advisory," Reavey said, which would mean the company knew of attacks on 5 July, nearly a month after IBM said attacks had started. "Once we saw the attacks, we took a look at the current status [of our work] and what's being attacked, [and] put things on a fast track."
On a Microsoft blog announcing the security updates planned for release next week, an MSRC spokesman said, "...our engineering teams have been working around the clock to produce an update."
Microsoft also denied that vulnerability details had leaked to hackers at some point during the last 16-18 months, perhaps through the Microsoft Active Protection Program (MAPP), a programme that gives security software companies early information on bugs. "Microsoft did not share any information with MAPP partners about the reported Video ActiveX Control vulnerability until immediately before the advisory posted," a company spokesman said.
Hackers are exploiting the ActiveX vulnerability by getting users to visit malicious sites, or planting drive-by attack code on legitimate sites. The number of compromised sites serving up the malware to IE6 and IE7 users has skyrocketed, and number in the millions, according to ScanSafe.
At some point, Microsoft will release a true patch for the problem, Reavey said. He declined to say whether that patch would be delivered "out-of-cycle" - outside the normal monthly update schedule - when it is ready, however.