On Mon, Feb 04, 2019 at 10:55:18PM -0800, Christoph Hellwig wrote: > On Mon, Feb 04, 2019 at 04:26:38PM -0800, Darrick J. Wong wrote: > > I decided to leave the unlinked counter so that we could have a > > debugging check. I will make it more explicit that anyone accessing the > > counter needs to hold the AGI buffer lock or otherwise assured that > > there aren't any other threads. > > Don'we already have a debug check by the fact that the rhashtable > destroy function throws an assert if it finds anything? Yes. If we're maintaining the backrefs correctly then we should never have any left over and presumably we shouldn't ever need to fall back to the bucket walk. Ok, dropped. --D
