attached is a 0.46-r1 ebuild which changes the random device to /dev/urandom if the urandom USE flag is set. It does not change the behavior of the existing "fallback to urandom" patch. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Created attachment 74761 [details] r1 ebuild with USE=urandom support
i dont see much benefit in this
The benefit is that when you know /dev/random does not work, such as on an embedded system, you can skip the extra-long timeout and go straight to /dev/urandom. The fallback approach in the ebuild is nice, but why not let users have the control if they know they need it? And why would the dropbear maintainer implement the feature if it wasn't supposed to be used?
i was going to fix the fallback patch to not delay
How would that work? Doesn't the fallback have to wait till /dev/random times out before trying urandom?
read the patch ... i would merely remove the call to sleep()