On Wed, Jul 25, 2012 at 1:25 PM, Ted Unangst <t...@tedunangst.com> wrote: > tc_setclock - doesn't seem to do much itself. What are the > consquences of timehands->th_offset getting raced?
Racing calls to {micro,nano,bin}time() can get bogus times (e.g.,
interrupts or even other userspace threads that call clock_gettime(),
since it's an unlocked syscall). It should probably do the same
th_generation dance as tc_windup().
