DNS bug tattler not the first to guess flaw details

Two weeks of silence helped, says researcher who found critical flaw

The researcher whose speculation led to an early disclosure of information about a critical flaw in the Domain Name System (DNS), the Internet's traffic cop, wasn't the first to come close to the truth, said the security expert who found the bug and organized a massive patching effort.

"Halvar [Flake] was not the first, not even the tenth," said Dan Kaminsky, director of penetration testing at Seattle-based IOActive and the researcher who uncovered the DNS flaw early this year and helped coordinate a multi-vendor patching process that kicked off two weeks ago. "A lot of other people figured this out first."

On Monday, Flake, the hacker moniker of Thomas Dullien, CEO of the German security company Zynamics, took a stab at the flaw and posted his best guesses about its details and how it might be exploited. Later on Monday, Flake's speculations were confirmed by Matasano Security, a security consultancy that included at least one researcher who had been briefed on the bug by Kaminsky several days later.

"I asked [others who had guessed the details] to hold off until Black Hat," said Kaminsky late Tuesday. "I really have to express my appreciation to them. The security community did not speak with one voice. Many [researchers] realized the importance of this."

When Kaminsky first announced the bug, he said he would provide technical details on August 7 at the Black Hat security conference in Las Vegas. At the time, he said he wanted to give corporate and organization administrators a month to patch before he got specific about the bug.

Kaminsky seemed to bear no ill toward Flake for posting details of the DNS flaw. "I'm not going to say it was irresponsible," said Kaminsky of Flake. "It's his blog and he can say what he wants on it. I can only ask [that others hold information], I can't demand. My only regret is that I didn't have more of an opportunity to talk to him before he posted."

Kaminsky said he and Flake had traded e-mails before Flake went public.

Immediately after Flake's post and Matasano's confirmation, Kaminsky urged administrators responsible for DNS servers to patch immediately, a recommendation he repeated Tuesday. "There's definitely [more of] an increased risk than there was two days ago," he said. "I would advocate doing what you can [to patch now]. I tried to give you as much time as I could. I knew that 30 days was not going to be enough, but we didn't get them. But 13 was better than zero."

He was referring to the nearly two weeks between Flake's posting and availability of patches from Cisco Systems, the Internet Systems Consortium and Microsoft.

The flaw, he said, is serious. How serious? "Who do you want to be able to send e-mail to?" he asked as answer. "This is absolutely serious. Do you want to see the sites you expect? Or do you want to put people in the position where [attackers and identity thieves] are sending everyone to [servers in] China? Then patch."

With details of the DNS bug out of the bag, other researchers have said exploits won't be long in coming. HD Moore, creator of the Metasploit penetration testing framework, told the IDG News Service yesterday that attack code will probably appear soon, and could present big problems for months to come.

Page Break

Kaminsky said the short-term impact from the bug going public is that some administrators will have to take down their networks unexpectedly in order to patch.

But he had little sympathy for those who had brushed off his earlier warnings. "Some people said,'I can't justify doing anything without knowledge of the exploit,'" Kaminsky said. "To those people, all I can say is,'So, do you feel safer now?'"

Kaminsky also had words for security researchers who don't understand how difficult it is to organize a large-scale patching effort, or even how much time it takes to patch a widespread flaw throughout a corporation's network. "Our job as security researchers isn't just to find bugs, but to get them fixed," he argued. "Breaking stuff is easy. Who has the easier job? The one who finds the bug or the one who helps fix it? The hacker has the easier job here.

"This isn't a trivial patch. It's going to take a huge amount of effort. A lot of organizations are going to be rough on their IT staffs to get this done."

He was encouraged, however, by the progress that had been made in patching the bug, even in just 13 days. "There are a lot of people who have patched, not a majority, but 30 per cent to 40 per cent of those tested [using a widget on Kaminsky's site were safe. Those are enormous numbers for such a short time. Is that everyone? No," said Kaminsky. "But those that started [when patches were first available] have a two-week head start over those who waited."

Kaminsky, however, wasn't ready to get specific about what he would have done differently. Those lessons, he said, would have to wait until after Black Hat. "I did everything I could for my customers," he concluded. "I have taken a lot of dings, but I would do it again."