Chris Jones <cjones@rupert.honors.montana.edu> writes:

> So, what's the current status of /dev/random?  I know that the kernel
> config files claim it's still experimental; why is this?  Does it
> appear to work, but it hasn't been verified by a statistician?  Does
> it work, but produce insufficiently random numbers?  Does it crash the
> machine?

There was some controversy as to how good it really is.  Core has
dubbed it ready for the Experimental tag to be removed, and it will
soon be turned on for all ports.

However, before that happens, I want to change the entropy gathering
method a bit, and whack on the code a bit.  It'll be ready for 1.4,
and not marked as "experimental" anymore.

However, serial port devices (on i386 at least) cannot be used for
timing information yet...  This is because serial ports run at
splserial, which happens to be higher priority than splhigh where the
entropy internals run.

--Michael