This code has had an extensive rewrite and a good series of reviews, both by the author and other parties. This means a lot of code has been simplified. Pluggable structures for high-rate entropy generators are available, and it is most definitely not the case that /dev/random can be driven by only a hardware souce any more. This has been designed out of the device. Hardware sources are stirred into the CSPRNG (Yarrow, Fortuna) like any other entropy source. Pluggable modules may be written by third parties for additional sources.
The harvesting structures and consequently the locking have been simplified. Entropy harvesting is done in a more general way (the documentation for this will follow). There is some GREAT entropy to be had in the UMA allocator, but it is disabled for now as messing with that is likely to annoy many people.
The venerable (but effective) Yarrow algorithm, which is no longer supported by its authors now has an alternative, Fortuna. For now, Yarrow is retained as the default algorithm, but this may be changed using a kernel option. It is intended to make Fortuna the default algorithm for 11.0. Interested parties are encouraged to read ISBN 978-0-470-47424-2 "Cryptography Engineering" By Ferguson, Schneier and Kohno for Fortuna's gory details. Heck, read it anyway.
Many thanks to Arthur Mesh who did early grunt work, and who got caught in the crossfire rather more than he deserved to.
My thanks also to folks who helped me thresh this out on whiteboards and in the odd "Hallway track", or otherwise.
My Nomex pants are on. Let the feedback commence!
Reviewed by: trasz,des(partial),imp(partial?),rwatson(partial?)
Approved by: so(des)
1) Clean up namespace; only use "Yarrow" where it is Yarrow-specific
or close enough to the Yarrow algorithm. For the rest use a neutral
name.
2) Tidy up headers; put private stuff in private places. More could
be done here.
3) Streamline the hashing/encryption; no need for a 256-bit counter;
128 bits will last for long enough.
There are bits of debug code lying around; these will be removed
at a later stage.
can more easily be used INSTEAD OF the hard-working Yarrow.
The only hardware source used at this point is the one inside
the VIA C3 Nehemiah (Stepping 3 and above) CPU. More sources will
be added in due course. Contributions welcome!
o Each source gets its own queue, which is a FIFO, not a ring buffer.
The FIFOs are implemented with the sys/queue.h macros. The separation
is so that a low entropy/high rate source can't swamp the harvester
with low-grade entropy and destroy the reseeds.
o Each FIFO is limited to 256 (set as a macro, so adjustable) events
queueable. Full FIFOs are ignored by the harvester. This is to
prevent memory wastage, and helps to keep the kernel thread CPU
usage within reasonable limits.
o There is no need to break up the event harvesting into ${burst}
sized chunks, so retire that feature.
o Break the device away from its roots with the memory device, and
allow it to get its major number automagically.
one out of a block cipher. This has 2 advantages:
1) The code is _much_ simpler
2) We aren't committing our security to one algorithm (much as we
may think we trust AES).
While I'm here, make an explicit reseed do a slow reseed instead
of a fast; this is in line with what the original paper suggested.
o Separate the kernel stuff from the Yarrow algorithm. Yarrow is now
well contained in one source file and one header.
o Replace the Blowfish-based crypto routines with Rijndael-based ones.
(Rijndael is the new AES algorithm). The huge improvement in
Rijndael's key-agility over Blowfish means that this is an
extremely dramatic improvement in speed, and makes a heck of
a difference in its (lack of) CPU load.
o Clean up the sysctl's. At BDE's prompting, I have gone back to
static sysctls.
o Bug fixes. The streamlining of the crypto stuff enabled me to
find and fix some bugs. DES also found a bug in the reseed routine
which is fixed.
o Change the way reseeds clear "used" entropy. Previously, only the
source(s) that caused a reseed were cleared. Now all sources in the
relevant pool(s) are cleared.
o Code tidy-up. Mostly to make it (nearly) 80-column compliant.
call instead.
This makes a pretty dramatic difference to the amount of work that
the harvester needs to do - it is much friendlier on the system.
(80386 and 80486 class machines will notice little, as the new
get_cyclecounter() call is a wrapper round nanotime(9) for them).
now in dirs called sys/*/random/ instead of sys/*/randomdev/*.
Introduce blocking, but only at startup; the random device will
block until the first reseed happens to prevent clients from
using untrustworthy output.
Provide a read_random() call for the rest of the kernel so that
the entropy device does not need to be present. This means that
things like IPX no longer need to have "device random" hardcoded
into thir kernel config. The downside is that read_random() will
provide very poor output until the entropy device is loaded and
reseeded. It is recommended that developers do NOT use the
read_random() call; instead, they should use arc4random() which
internally uses read_random().
Clean up the mutex and locking code a bit; this makes it possible
to unload the module again.
from many folk.
o The reseed process is now a kthread. With SMPng, kthreads are
pre-emptive, so the annoying jerkiness of the mouse is gone.
o The data structures are protected by mutexes now, not splfoo()/splx().
o The cryptographic routines are broken out into their own subroutines.
this facilitates review, and possible replacement if that is ever
found necessary.
Thanks to: kris, green, peter, jasone, grog, jhb
Forgotten to thank: You know who you are; no offense intended.
