Microsoft confirms serious IIS bug, downplays threat

Microsoft late Monday confirmed that its Internet Information Services (IIS) Web-server software contains a vulnerability that could let attackers steal data, but downplayed the threat.

"An attacker could exploit the vulnerability by creating a specially crafted HTTP request to a Web site that requires authentication, and thereby gain unauthorized access to protected resources," Microsoft said in a security advisory issued Monday night.

"[But] only a specific IIS configuration is at risk from this vulnerability," Jonathan Ness, an engineer with the Microsoft Security Response Center (MSRC), said in a post to the center's blog.

Earlier in the day, security organizations, including Cisco and the U.S. Computer Emergency Response Team (US-CERT) had warned that IIS 6 harbored a bug that a researcher claimed could be used to both view and upload files to Web servers.

According to Microsoft, the flaw affects IIS 6 servers where WebDAV (Web-based Distributed Authoring and Versioning), a set of extensions to HTTP used to share documents over the Web. WebDAV is also used in Microsoft Exchange 2003 to access inboxes through a browser.

Microsoft also confirmed that the older IIS 5 and IIS 5.1 software is vulnerable; The newer IIS 7, which debuted alongside Windows Vista and is included in Windows Server 2008, is not affected, however.

The vulnerability was revealed last week in a message by security researcher Nikolaos Rangos on the Full Disclosure mailing list. Although Rangos said the bug could be used to upload potentially malicious files, Belgian researcher Thierry Zoller said there was no way for an attacker to actually run malware planted on the server.

Ness echoed Zoller, with the caveat that Microsoft is still looking at the bug. "What we have found is that the IIS installer applies an NTFS access control entry to explicitly deny write access to the anonymous account (IUSR_[MachineName]) in wwwroot and subdirectories that inherit wwwroot's ACL," he said. "So in the default case, this vulnerability will not allow a malicious attacker to upload or modify Web pages."

He also ticked off four criteria that must be met to put a server at risk, and noted that "this vulnerability is primarily an information disclosure threat."

Rangos' bug can be traced to 2001, the year that the "Code Red" worm slowed Windows-based networks to a crawl, said Zoller, who noted that eight years ago Microsoft patched a path traversal bug in May 2001. "Its resemblance to the IIS Unicode flaw from 2001 was so similar that my jaw first dropped," he said in a blog entry last Saturday. "The bug discovered by Rangos seems to suffer from a similar logic mistake [as MS01-026. Later that year, Microsoft patched other IIS bugs, including the one exploited by Code Red.

This newest flaw, however, is not related to the Code Red vulnerability.

Microsoft's Ness outlined several workarounds that users could take until a patch was available, including disabling WebDAV, in IIE 5, 5.1 and 6. The company did not explicitly promise a patch, but its advisory included boilerplate language -- "Microsoft will take the appropriate action to help protect our customers -- that typically indicates a fix is forthcoming.

For more enterprise computing news, visit Computerworld. Story copyright Computerworld, Inc.