On Sun, 22 May 2011 20:12:04 -0500 Steve French <smfrench@xxxxxxxxx> wrote: > On Sun, May 22, 2011 at 7:17 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote: > > On Sun, 22 May 2011 07:59:38 -0400 > > Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote: > > > >> On Sun, May 22, 2011 at 07:55:24AM -0400, Jeff Layton wrote: > >> > This will never work properly with CIFS, as the protocol has no ability > >> > whatsoever for looking up files by filehandle. It *might* be possible to > >> > eventually do this with SMB2, but that remains to be seen. > >> > > >> > For now, it just plain doesn't work. Mark it BROKEN to discourage > >> > distros from enabling it. > >> > >> It's actually dead code at this point - fs/nfsd/vfs.c:check_export() > >> refuses to export a filesystem that does not have a fh_to_dentry > >> operation, which cifs doesn't provide. > >> > >> IMHO it's best to just remove the option and all surrounding code. > >> > > > > I agree and proposed a patch to do that a month or two ago. Steve > > NAK'ed it for reasons that I don't quite understand. > > We were waiting on clarification on NTCreateX open by inode number flag. > It's clarified -- it doesn't do anything for directories so it doesn't help. > We have had multiple requests to be able to do limited > export of cifs data via nfs (for example for backup to/from > OS with only one of the protocols supported). NFS clients > don't make this easy - because some don't handle ESTALE > by revalidating their dentries. > > For nfs v3 clients that can't handle ESTALE, we are probably > stuck with having to wait for Samba to implement the flag. > Flag? > I did want to see JRA's opinion and also query if other > NFS clients (other than Linux) can handle ESTALE > (by e.g. revalidating the dentry and repeating the operation). > The client can't always revalidate the dentry. For instance, suppose you open a file or directory and then it gets renamed or removed on the server. The client may have no idea what the new path actually *is* in order to revalidate it. The fact that CIFS has no lookup by filehandle makes this whole idea dead in the water. SMB2 might be able to do something, but I'm not yet convinced of that. Either way, to do this will require more than the stubs in place now. I think this code either needs to be removed until you come up with a design that overcomes these problems, or marked broken (like this patch does). -- Jeff Layton <jlayton@xxxxxxxxxx> -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
