On 06/01/2012 03:22 PM, Adam Williamson wrote:
On Fri, 2012-06-01 at 15:14 -0400, Gerry Reno wrote:
> I just read through the MS docs on SecureBoot and this is the biggest Rube-Goldberg
machine.
>
> I could not think of a nastier solution to a problem than what they've dreamt up
here.
>
>
> The whole problem they are trying to solve is that of booting only known-good code.
>
>
> That would be much easier accomplished by having the OS reside on a read-only device
that could only be written to by
> the user actively using hardware to enable the write during installation.
>
> That would create a system where there was no possible programmatic means of
corrupting the OS during normal operation.
>
> No signatures, no crypto-databases, or other SecureBoot gobbledy-gook needed.
>
>
> To implement this would require only that new systems support two drives, one with
controllable-by-user
> read-write-controller interface for storing the OS.
>
> Forensic firms have been using these types of read-write controllable drive
interfaces for years. Hardware already exists.
What is your practical point?
My practical point is that Microsoft chose this particular solution not as the best way to
solve the issue of booting
known-good code but as a way of impacting Linux and it whole concept of software
freedoms.
I don't think anybody in the Linux community should be supporting this SecureBoot
"solution" in any way, shape or form.
.