Darren J Moffat wrote: > Dan McDonald wrote: >> Let's say I have an app that uses /dev/random and can possibly >> contribute >> data to the pool. >> >> In my app, remote peers generate on their own a 64-bit SPI value. >> Remote >> peers also come from different remote IP addresses, and sometimes >> different >> remote ports. >> >> I'm curious if I stir this in if it will overwhelm /dev/random or if its >> entropy sources more than dominate anything I can contribute. Are there >> guidelines for good *writers* into /dev/random? > > You can write as much as you like to /dev/random providing you have > permission to write to the device. We take what you give us with a > "grain of salt" and only use as much of it as we can deal with. >
Darren is correct when he says "grain of salt". The effect of seeding is very small. Applications really don't need to seed the opensolaris /dev/random in order to get high quality numbers. Of course this might not be the case for other implementations of /dev/random on other operating systems. Mark
