[linux-cifs-client] Re: jeff patches

[Steve French]

comments on patch series follow.  Remainder are merged now.

On Tue, Sep 23, 2008 at 12:50 PM, Steve French <smfrench at gmail.com> wrote:
1)        cifs: make cifs_rename handle -EACCES errors      looks fine but has
dependencies
2)        cifs: update DOS attributes in cifsInode if we successf ...
    looks fine but has dependencies
3)        cifs: clean up error handling in cifs_unlink
4)        cifs: undo changes in cifs_rename_pending_delete if ...
I think this isn't right.   In any case we should set delete pending
even if we can't rename here:
        rc = CIFSSMBRenameOpenFile(xid, tcon, netfid, NULL, cifs_sb->local_nls,
                                   cifs_sb->mnt_cifs_flags &
                                            CIFS_MOUNT_MAP_SPECIAL_CHR);
-       if (rc != 0)
-               goto out_close;
+       if (rc != 0) {
+               rc = -ETXTBSY;
+               goto undo_setattr;
+       }

        /* set DELETE_ON_CLOSE */
        rc = CIFSSMBSetFileDisposition(xid, tcon, true, netfid, current->tgid);
+       if (rc != 0)
+               goto undo_rename;

5)        cifs: add function to set file disposition
     merged
6)        cifs: move rename and delete-on-close logic into helper ...
     merged
7)       cifs: fix busy-file renames and refactor cifs_rename ...
     Question to resolve still on why forbid cross directory renames?
8)        cifs: remove NULL termination from rename target in ...
    looks fine but waiting for answer from ms
9)        cifs: add constants for string lengths of keynames ...
   fixed up and merged
10)        cifs: clean up upcall handling for dns_resolver keys
                                  clarify or find similar usage of
generic fields being overloaded by your patch x.[0] x.[1] (are they
 described somewhere?)
11)        cifs: no need to use rcu_assign_pointer on immutable ...
                                  seems like it doesn't matter
12)        cifs: explicitly revoke SPNEGO key after session setup     the revoke
 vs. put - who cleans up question needs to be doublechecked
                                  at first glance it does not look
like revoke is a superset of put
13)        cifs: have find_writeable_file prefer filehandles opene ...
     merged




-- 
Thanks,

Steve