z/OS client restrictions

Because z/OS Unix does not have a unique Base Mask ACL entry:
- The z/OS NFS Client uses the Base Group ACL entry permission value for calculating the Mask ACL entries sent to NFS Servers.
- The z/OS NFS Client fails 'get attribute' requests containing Mask ACL entries which do not match the Base Group ACL entry. The failure returns an ENODEV error.
- Unlike Sun, the z/OS NFS Client set both the Group and Mask ACL entries when a 'chmod' is issued for a file/directory which has Extended ACLs. If only the Group or Mask ACL entry were changed, it would no longer be possible to display or change the ACLs for this object since the Group and Mask entries would be different.
Because z/OS Unix does not have unique Default Base ACL entries:
- The z/OS NFS Client uses the Access Base ACL entries for calculating the Default Base ACL entries sent to NFS Servers (that is, the Access and Default Base ACL entries will always match).
- The z/OS NFS Client fails get attribute requests containing Default Base ACL entries that do not match the Access Base ACL entries. The failure returns an ENODEV error.
Because non-z/OS platforms do not have separate File and Directory Default ACLs:
- The z/OS NFS Client uses the File Default ACL entries for the NFS Default ACL entries sent to non-z/OS NFS Servers. The entries are used as both File and Directory Defaults on the non-z/OS platforms.
  The z/OS NFS Client checks any Dir Default ACL set requests to verify that they match the existing NFS Default ACL entries. If the two lists do not precisely match (same entries with same permissions), the set request fails with an ENODEV error. This requires that the File Default ACL entries must be set before the Directory Default ACL entries.
- Ideally, Directory Default ACL entry set requests should not be issued in this case.
- The z/OS NFS Client uses the NFS Default ACL entries sent by a non-z/OS Server for both File and Directory Default ACL get requests.
The z/OS NFS Client supports the File and Directory Default ACL entries as unique entries when communicating with a z/OS NFS Server. There are no request order or permission value restrictions for this case.