On Fri, Jun 23, 2017 at 09:01:46AM +0200, Michael Haggerty wrote: > Change `repack_without_refs()` to expect the packed-refs lock to be > held already, and not to release the lock before returning. Change the > callers to deal with lock management. > > This change makes it possible for callers to hold the packed-refs lock > for a longer span of time, a possibility that will eventually make it > possible to fix some longstanding races. This is the sort of clue I was thinking about in my last email. :) > The only semantic change here is that `repack_without_refs()` used to > forgot to release the lock in the `if (!removed)` exit path. That > omission is now fixed. s/used to forgot/previously forgot/ or similar? > @@ -731,14 +717,12 @@ int repack_without_refs(struct ref_store *ref_store, > * All packed entries disappeared while we were > * acquiring the lock. > */ > - rollback_packed_refs(refs); > + clear_packed_ref_cache(refs); > return 0; And this is the reason for the earlier "you should be able to clear the packed ref cache without holding the lock" commit, I guess. Makes sense. -Peff
