News for hackers. Stuff that matters.
Re:/dev/random – a new approach by aortega on Monday April 25, 2016 @03:02AM (#11562845)
>Kernel CSPRNGs do need continually updated entropy pools. But not because their original entropy is depleted! Rather: you want your RNG to be updated so that its "key" changes regularly, so that if your system is compromised transiently, you'll recover.

It's very unlikely that a compromise will be temporary. This situation "transient compromise" will never happen. Moreover, once a system is compromised, by definition you cannot trust it until replacing it from a known-good source or backup, at which point, you just start the CSRNG with new entropy.

No system should need or even attempt to recover from a transient compromise, you are just causing a false feeling of security.

Re:/dev/random – a new approach by ibotty on Monday April 25, 2016 @03:23AM (#11562893)
You don't necessarily know your rng is being attacked.

Rekeying frequently makes passive attacks way harder, so it makes sense to rekey as a often as you can.

Re:/dev/random – a new approach by aortega on Monday April 25, 2016 @05:59PM (#11567716)
What would be, for example, a passive attack against a CS-RNG?
Re:/dev/random – a new approach by Natanael_L on Tuesday April 26, 2016 @07:59AM (#11571061)
Sidechannel leaks against mobile devices.