IPredator complains that we call time() too much when running over 500Mbps
We are constantly testing kernel 3.x vs. the latest 2.6, but noticed that with kernel 3.x beyond 500MBit Tor would spend massive amounts of CPU time on calling time(). On kernel 2.6 we use kernel.vsyscall64=2 which minimizes the impact of constantly doing that.
nickm, is this still an issue? This would be possible to simulate using chutney, right?