Trust Gets Buggy

This would be funny if it wasn't so sad. Microsoft has in the past three months declared that two security bugs are, in fact, features. The most recent involves IIS, the earlier one involves Word 2007. Apparently nobody at Microsoft realizes that the "It's not a bug, it's a feature!" should only be used as a punch line to a joke.

In the case of the Word bug, there is at least the consolation that Word is hardly an enterprise server application, as pointed out by Microsoft's David LeBlanc. In his analysis, crashing an application, in a way that you are "reasonably certain" it won't be exploitable, is a good thing. And he points out that you need to be absolutely certain you can safely return to code execution after catching an exception, and that triggering a crash may be the best way to do that.

Of course, in this case, the flaw is in the code that, um, loads a document. If the .docx parser, which is for a new file format, has so much code in it that wasn't designed for exception handling, it should be clear to even the most casual observer that there can be no reasonable expectation of a lack of exploitable bugs. This is really, really sad.

For IIS, we are in a different realm entirely. IIS is billed as an enterprise class Web server, suitable for all sorts of Web applications. The bug discovered finds the hit highlighting feature of IIS 5 relies only on ACL-based security, not on any sort of Web-server mechanisms (NTLM/basic authentication, IP restrictions).

The proposed solution? Upgrade to IIS 6 or better. Of course such an upgrade would likely involve an upgrade of server OS, testing of all Web server code on the new version, and a tremendous amount of pain and suffering for IT staff.

Naturally, a reasonable argument can be made that this work is inevitable, and that since IIS 6 has been out for several years now, it's about time to upgrade anyway.

The real problem here, however, isn't in the proposed solution. If Microsoft chooses to stop supporting IIS 5, that is within its rights. The problem is the attitude that a design flaw cannot, by definition, be a security bug.

Let's walk further down this same path. If I intentionally store passwords in plain text, or display a credit card number on the screen, or store users' Social Security numbers in cookies without asking first, then these are somehow not security flaws.

Hasn't Microsoft learned yet? For all the hype about Trustworthy Computing, the company sure don't seem like it understands much, or that the people providing the software have to be worthy of our trust.

— Nathan Spande has implemented security in medical systems during the dotcom boom and bust, and suffered through federal government security implementations. Special to Dark Reading.