Much ado about signed drivers, and yet…

Today, I was installing a Srv03 x64 VM for testing purposes. This is something I’ve done countless times before – install/reinstall/blow away Windows VMs for testing purposes with the standard Windows setup, nothing all that interesting. However, I ran into something extremely bizzare this time around:

Windows Server 2003 x64 Setup - Unsigned Driver Warning

The really weird thing here is that this VM was completely clean; brand new, empty hard disk which had only been just now connected to standard installation media (no third party OEM drivers slipstreamed onto the media or anything either).

Apparently, there’s an unsigned “in-box” driver lurking on the Srv03 x64 installation discs. Nice. The only thing that I can think of that was different about this VM from all the others I’ve made is that I created a serial port (redirected to a named pipe as usual) in the VM before setup, instead of after, and that somehow Windows might have thought that I had one of those old-style serial-port-controlled UPS batteries connected to the box. Still, I’ve been around enough Srv03 installs on “real metal” to know that it’s weird for this to happen; I’ve never seen it on another install, VM or physical hardware.

Given the fuss that Microsoft makes about signed drivers, seeing what appears to be an in-box driver that isn’t signed is, to say the least, amusing. (Note that there are two types of signing; cat file signing, to sign the installation package – this is what suppresses the unsigned driver popup – and driver binary signing, which allows a driver to load on an x64 system but still generates the unsigned driver popup if you install the driver via an INF. I can only assume that for whatever reason, this driver doesn’t have a valid catalog signature, or it wouldn’t load at all. I’ll see if I can confirm this when the VM finishes installing, though…)

Leave a Reply