Adding oplocks to SMB2 == changing the share_mode entry struct.

[Jeremy Allison]

Volker, Metze, and all,

Adding oplocks to the SMB2 server code (which I'm looking at
at the moment) will mean changing the definition of the
share_mode_entry struct (and associated data like the
linearized oplock break message passed between smbds)
by changing the op_mid entry from uint16_t to uint64_t
(the size of an SMB2 message id value).

This will affect clusters in that it will mean all
nodes of a cluster will have to be upgraded at once
(cluster shutdown then restart).

I'd prefer to do it this way rather than adding real
uglyness like a mapping table of SMB2 mids -> SMB1
mids.

Can you give me some feedback on your thoughts on
this ? I'm planning to start putting this code in
place asap.

Thanks !

Jeremy.