Brian Gladman, a UK security consultant who has worked with a number of companies on issues surrounding the TCPA (Trusted Platform Computing Alliance) has written anexcellent poston the subject at cryptome.org. His post is a response to an earlier post by Ross Anderson, which I linked up and wrote about in thisnews post. Gladman, while not exactly excited over the TCPA, takes a more moderate tone and tries to look at the ways in which the initiative, if done properly, could actually benefit consumers. Of course, “properly” is the operant word here, and though Gladman has some very sensible ideas about what a proper implementation would look like he doesn’t seem to expect it to pan out without a lot of pressure from consumers. So Gladman shares Anderson’s concerns about the TCPA’s likely negative impact on consumers, even though he still has some small hope that the TCPA could be done in a way that benefits all parties involved.
The main place where Gladman disagrees with Anderson is on whether or not it would be possible to build GPL software for a TCPA platform. I would quote some of Gladman’s argument, but I’m afraid that people might respond to the quote and not read his entire post. Also, I’m not an expert on the ins and outs of the GPL, so I can’t actually evaluate Gladman’s claims in this area. If anyone who’s more informed would like to respond to this aspect of Gladman’s argument, please post in the discussion thread.
On a related note, I was digging through the memory subsystem section of Hennessy and Patterson (chapter 5) in preparation for the forthcoming sequel to my “Understanding CPU Caching and Performance” article, and I came across an interesting section on an elaborate, arcane, and mostly unused(?) memory protection mechanism that Intel introduced into its processor line way back with the 8086. This scheme was designed to make it as difficult as possible for any part of the system to run malicious code, and it offers a variety of protections and safeguards that are intended to prevent various types of attacks and security breaches. Hennessy and Patterson regarded this memory protection scheme as overkill for a PC multiuser OS, and they wondered if it would ever be fully used. Their almost prescient (for 1996) conclusion to the section is worth quoting in its entirety:
