On Tue, 25 Jun 2013 01:42:44 -0500 Steve French <[email protected]> wrote:
> Looking at this issue at the plugfest this week - the recommendation > is to add a check for ATTR_REPARSE > since this is usually set (for the Windows case) for DFS links (DFS > links are one case of reparse points - but the performance hit for > revalidating all reparse points is not bad). Tomorrow we can check > for the Samba case in more detail to see how the scenario looks and > whether we can workaround it. > > We did find another workaround which could be checked, but decided > against using that one in preference to simply checking for the > reparse attribute. In the Windows case since you don't create > subdirectories under a DFS link on the server side - the creation time > and last write time of the directory will match for the case of a DFS > link returned by QueryDir or FindFirst (so we also could have refused > to cache directories whose creation time and last write time match - > which would also mean that we don't cache empty subdirectories which > is also a trivial perf hit to not cache those - it would be easy > enough to refuse to cache those directories too if you want). > Sounds reasonable -- you should probably post the patch you've already merged into your for-next branch so others can comment and review though... -- Jeff Layton <[email protected]> -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
