On Wed, Jun 1, 2016 at 10:37 AM, Theo de Raadt <[email protected]> wrote: >> Mark Kettenis wrote: >> > I sympathise with the idea, but your implementation would still print >> > "funny" times if your machine had been up for more than a day. >> >> The perils of rebooting. I tested with an hour, then bumped to a day for >> margin. But really, I think even up to ten years would work. Any "date" from >> the 70s is far more likely to be an interval than a timestamp, no? > > The purpose of ktrace is to help debugging. > > As it stands right now, it is very unreadable. It is difficult to > verify timestamps.
I think it would be better to include the clock_t value in the timespec struct tracing, then branch on that. Philip Guenther
