>
> > > 1. You create a problem for in kernel users of random numbers.
> > > 2. You forgo the benefit of randomness by concurrent access to
> > > /dev/urandom 3. You will not benefit from hardware random number
> > > generators as easily.
> >
> > You lost me. The kernel of course has "client" access to the internal
> > pool. And since the userspace reads from /dev/random, it benefits
>
> The kernel users of random numbers may be unable to block.
> Thus the kernel has to have a PRNG anyway.
> You may as well export it.
>
> > from HRNG just the same way it does now. Point 2 is somewhat obscure
> > to me. The kernel has only one observer to deal with, in theory.
>
> In theory. In practice what goes out through eg. the network is
> most important. Additional accesses to a PRNG bitstream unknown
> outside make it harder to predict the bitstream.
Not at all. Let me (one process) read 1MB from /dev/urandom,
and analyze it. If I can break SHA-1, I'm able to predict *future*
/dev/urandom output, expecially if I keep draining bits from /dev/random.
And even if it was not the case, you don't necessary need to read
the PRNG output *in order* to be able to guess it. Every N bits you
read, you learn more about its internal state.
Despite that, you tells you I'm not able to capture outgoing network
packets as well?
.TM.
--
____/ ____/ /
/ / / Marco Colombo
___/ ___ / / Technical Manager
/ / / ESI s.r.l.
_____/ _____/ _/ Colombo@ESI.it
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/