Simon Burge <sim...@netbsd.org> wrote: > > > > Simplify tmpfs_itimes() and use vfs_timestamp(). [ ... ] > > Was changing from getnanotime() to effectively nanotime() (via > vfs_timestamp()) deliberate? The original intention of using > getnanotime() for filesystem timestamps was that having a "perfect" > timestamp was less important than the time taken to obtain the > timestamp itself.
Yes, deliberate. It is consistent with other *_itimes() routines. I doubt that such performance difference matters in this place. However, if we want to move to getnanotime(), then it should be changed in vfs_timestamp(). -- Mindaugas