News for hackers. Stuff that matters.
Re:/dev/random – a new approach by mikepurvis on Sunday April 24, 2016 @05:29PM (#11561443)
The linked PDF: http://www.chronox.de/lrng/doc/lrng.pdf

Page 4: "The new approach implements the modeling of a slow noise source within the LRNG based on the timing of interrupts and allowing other, even fast operating noise sources to contribute entropy. As discussed above for the legacy /dev/random, only the high-resolution time stamps deliver entropy for hardware events and other information processed by the legacy /dev/random implementation hardly have any entropy. This identification is used as a basis that solely the timing of interrupts is used."

Re:/dev/random – a new approach by acqq on Sunday April 24, 2016 @05:56PM (#11561544)
"The cryptographic processing of the entropic data is implemented with a well-known and well-studied deterministic random number generator: an SP800-90A DRBG as defined in [2] – naturally excluding the NSA-sponsored Dual-EC DRBG."

"The concept of the LRNG allows the selection of the used DRBG and the backend cipher implementations at compile time."

Is SP800-90A DRBG really state of the art?

According to Wikipedia "the publication contains the specification for four cryptographically secure pseudorandom number generators for use in cryptography." Without Dual-EC it's still three different DRBGs. How good are they? Is there anything better?

Re:/dev/random – a new approach by tptacek on Sunday April 24, 2016 @06:34PM (#11561676)
They're fine. They're sane, straightforward, relatively simple constructions based on hashes, HMAC, and ciphers running in CTR mode. You could make a CSPRNG more complicated, but there's not much reason to.

The stuff that extends the "state of the art" beyond SP800-90A tends to be system-specific and feature-based. For instance, the "estimators" in the LRNG are an extension to SP800-90A (and a bad idea). Rolling over seed files are another extension.

Re:/dev/random – a new approach by acqq on Sunday April 24, 2016 @06:39PM (#11561699)
> They're fine

Thanks. Does it matter which one of the three is then selected? Only one can be a default.

Re:/dev/random – a new approach by tptacek on Sunday April 24, 2016 @07:16PM (#11561793)
Meh. Throw a dart.