Messages in this thread | | | Date | Tue, 4 Feb 2014 19:23:25 +0000 | From | tytso@mit ... | Subject | Re: [RFC PATCH 0/5] CPU Jitter RNG |
| |
On Tue, Feb 04, 2014 at 11:06:04AM -0800, H. Peter Anvin wrote: > > The quantum noise sources there are in a system are generally two > independent clocks running against each other. However, independent > clocks are rare; instead, most clocks are in fact slaved against each > other using PLLs and similar structures.
One of the things that would be useful for us to understand is in general, where in a system we have independent clocks. For example, I think (correct me if I'm wrong), a 2.5" or 3.5" HDD has its own clock which is separate from the CPU/chipset. That is actually how and where we get any entropy; I am not at all convinced that we are getting any variation from "chaotic air turbulence in the HDD" --- that paper was published in 1994, and hard drive technologies have changed quite a bit since then, with extra layers of caching, track bufers, etc.
However, where a decade ago the ethernet card probably had its own independent clock crystal/oscillator, I'm going to guess that these days with SOC's and even on laptops, with ethernet device part of the chipset, it is probably being driven off the same master oscillator.
I wonder if there's anyway we can either figure out manually, or preferably, automatically at boot time, which devices actually have independent clock oscillators.
- Ted
| |