On Mon, 2019-04-15 at 12:11 -0400, Chuck Lever wrote: > > > I was about to try something like this. > So, the reason for that approach is that I believe we can currently leak ENOBUFS and possibly other errors in addition to EBADSLT through the same race. -- Trond Myklebust Linux NFS client maintainer, Hammerspace trond.myklebust@xxxxxxxxxxxxxxx
