We are experiencing undesired behavior using 'cp' and its siblings involving a NetApp filer qtree with the CIFS security model.
The CIFS security model on this type of system, or at least in our environment, allows Windows ACLs to control the permissions of directories/files. UNIX accounts map to AD accounts, so UNIX users can have varying permissions controlled by the Windows ACLs. Files in these areas look like this in UNIX, -rwxrwxrwx 1 username group 9 2007-10-31 10:24 file 'username' doesn't necessarily have 'rwx' to the file. It depends on what Windows ACLs are set. UNIX functions like chmod will fail, because changing of the permission bits is not allowed by the filer. This has not been a problem for everyday use by users who understand this security model we have implemented, but now we see new behavior with the plain-jane 'cp' command. The problem appears to be that 'cp' is trying to change the permissions on a new destination file before it copies over the contents of the file. /bin/cp file file2 /bin/cp: cannot create regular file `file2': Operation not permitted This creates the file, but does not transfer the contents... -rwxrwxrwx 1 username group 9 2007-10-31 10:32 file -rwxrwxrwx 1 username group 0 2007-10-31 10:32 file2 Performing another 'cp' will then successfully transfer the contents... -rwxrwxrwx 1 username group 9 2007-10-31 10:32 file -rwxrwxrwx 1 username group 9 2007-10-31 10:32 file2 What I assume is happening here is that 'cp' is trying to change the permissons of the file after it opens a new one for writing, and terminates when it cannot do so. This behavior occurs with coreutils versions 6.7 and 6.9, but not previous versions. It may be that the previous versions also tried to set permissions on a new file, but at least it didn't die when it couldn't do so. Before I start to investigate what has changed or develop a patch, I'm wondering if any developers know off-hand why the design in this area was changed, and if we have any hope of returning to previous behavior to support our use of our NetApp filer. Best Regards, Clarence Donath _______________________________________________ Bug-coreutils mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-coreutils
