Refilling /dev/random

[elsengoo]

I have a awk script for creating passwords and calculating statistical properties of the chosen type of passwords. It is working well on linux. But when I run it on the a780 (with busybox) /dev/random runs out of entropy quit fast and even after playing around a lot with several applications (button-press, joystick, internet-connection or even rebooting) I still can get only very, very few random bytes from this source.

How do I provide /dev/random with more entropy?

BTW. Floating point arithmethics in (busybox) awk results in completely wrong results. Is this a780 inherent or can it be solved?

[epinter]

you can use openssl library to generate random bytes ... It's better than /dev/random ...

I have a awk script for creating passwords and calculating statistical properties of the chosen type of passwords. It is working well on linux. But when I run it on the a780 (with busybox) /dev/random runs out of entropy quit fast and even after playing around a lot with several applications (button-press, joystick, internet-connection or even rebooting) I still can get only very, very few random bytes from this source.

How do I provide /dev/random with more entropy?

BTW. Floating point arithmethics in (busybox) awk results in completely wrong results. Is this a780 inherent or can it be solved?

[elsengoo]

That is not really a solution to my problem, since /dev/random is acting as a source of seeds for a pseudo-random-number -generator (PRNG). So using the PRNG of openssl would only shift the problem, since it needs a seed itself. Using local files to create a seed is an alternative option to using /dev/random that is already available, so the option -rand of openssl is also no solution.

So I prefer to get /dev/random refilled again. But how?

[numbski]

Having approached this problem about a year ago, I can tell you theory, but not a specific answer.

I would suggest finding the block device for the a780's microphone, and cat'ing that into /dev/random. That should be sufficiently random, and you should be able to get tons of random bits from that.

[epinter]

try non-block IO on /dev/random (I tested on solaris/sparc) or use /dev/urandom (works fine on E680i)...

[elsengoo]

@numbski:
Thanks. Now i only have to find out how to access the microphone block device...

@epinter:
IMHO non-block IO does not solve the problem of failing entropy on /dev/random. It only returns faster if no entropy is available, or did I miss something here?

I am using /dev/urandom as a temporary workaround now. But it is no real solution, since AFAIK it returns PRNG data, as soon as /dev/random gets out of entropy. And I want to use /dev/random as an alternative option to a PRNG as a "real random number generator".

Anyway, I am still interested to find out, how my a780 got the entropy in /dev/random initially before I started to drain it. It has not been refilled since several days now.

[numbski]

From what I follow, /dev/dsp and /dev/dsp16 are the audio block devices. If you search the forums I'm pretty sure you can find the answer you need.

[blackhawk]

- /dev/dsp = stereo playback device
- /dev/dsp16 = mono playback + record device = microphone device

[numbski]

Hey blackhawk! Thanks for filling in that blank, now maybe that will give me a clue on how to figure out the portaudio confusion in the iaxclient thread.

You tried the stuff I've uploaded?