Description of problem:
Under NetBSD hosts, when two processes acting on same file via same glusterfs fuse mount, second process is re-using the previously opened fd by first process. As a result fd opened by second process lacks requested flags when passed to glusterfs layer.

Version-Release number of selected component (if applicable):
master

How reproducible:
Always

Steps to Reproduce:
1. Setup a simple glusterfs distribute volume on a NetBSD 7 host.
2. Mount the volume under some mount point, say /mnt.
3. Switch off the performance tranlsators
4. Create a file under mount point, say /mnt/foo.
5. Compile and run lock.c with /mnt/foo as argument till it acquires a READ lock on it.
6. Attach gdb to brick process and put breakpoint at pl_writev.
7. From another session, compile and run write.c with /mnt/foo and perform the write.

Actual results:
fd->flags = 2 when breakpoint is hit

Expected results:
fd->flags = 6 when breakpoint is hit