Oplock problem: breaking our own lock

[Scott Moomaw]

I have found info on another lock-break failing scenario.  This involves a
single client accessing a Paradox database from a network share.  My
modification to smbd/open.c to increase debugging levels when a problem
occurs resulted in the log entries below.  In essence, the code is stuck
in an infinite loop because it cannot break its own oplock.

[2001/08/20 15:59:54, 5] smbd/open.c:open_mode_check(509)
  open_mode_check: oplock_request = 1, breaking oplock (1) on file
DIEBOLD/DATA/DATA.DB, dev = 740080, inode = 267927
[2001/08/20 15:59:54, 5] smbd/oplock.c:request_oplock_break(904)
  request_oplock_break: breaking our own oplock
[2001/08/20 15:59:54, 3] smbd/oplock.c:initial_break_processing(493)
  initial_break_processing: called for dev = 740080, inode = 267927 tv_sec
= 3b816bfd, tv_usec = 53657.
  Current oplocks_open (exclusive = 1, levelII = 0)
[2001/08/20 15:59:54, 3] smbd/oplock.c:initial_break_processing(510)
  initial_break_processing: cannot find open file with dev = 740080, inode
= 267927 allowing break to succeed.