Re: [edk2-devel] [Patch 0/2] Shadow microcode patch according to FIT microcode table.

["Laszlo Ersek" <lersek@redhat.com>]

On 01/08/20 10:42, Laszlo Ersek wrote:

> In closing: it seems short-sighted that the FIT specification placed a
> "naked" pointer at a fixed offset in flash, rather than a three-field
> structure consisting of:
> - a GUID,
> - preceded by a structure size,
> - preceded by the FIT pointer.
> 
> Because, using a GUID-ed approach, the chance to *incorrectly* deduce
> "yes FIT" would be 1 in (2^128) -- all 128-bit values except one magic
> value would indicate "no FIT". That's good.
> 
> Whereas, with the spec's current "naked pointer" approach, the chance to
> *correctly* deduce "yes FIT" is 3 in (2^64) -- all 64-bit values except
> three magic values indicate "yes FIT". Not good.

Sorry I messed up the first half of the last paragraph. I meant:

With the spec's current "naked pointer" approach, the chance to
incorrectly deduce "yes FIT" is ((2^64)-3) in (2^64). So an incorrect
decision is almost certain.

Thanks
Laszlo