> > Those workloads are broken garbage, and we should *not* use that kind > > of sh*t to decide on VFS internals. > > > > Sorry, I phrased it completely wrong. Thanks for clearing this up. I had just formulated my own reply but I'll happily delete it. :) > The workloads don't expect 1ns resolution. Yes, they don't. In the revert explanation I just used that number to illustrate the general ordering problem. The workload that surfaced the issue is just plain weird of course but it points to a more general ordering problem. > The workloads just compare timestamps of objects and expect some sane > not-before ordering rules. Yes. > If user visible timestamps are truncated to 100ns all is good. Yes.
