Lustre devel - Jan 2007 - [Bug 11415] 1.4.9.pre + sles 10: file locking broken

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11415

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|bugs@clusterfs.com          |green@clusterfs.com
             Status|NEW                         |ASSIGNED


Created an attachment (id=9283)
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
 --> (https://bugzilla.lustre.org/attachment.cgi?id=9283&action=view)
Add the locks to kernel''s list of locks.

Add the lock into list of kernel locks even for posix locks (if separate
->lock

method is present, if it is not (e.g. sles9 kernel, the kernel will do it by
itself).

This fixes problem with posix locks not being released on file close in 2.6.9+
kernels (including rhel4 and sles10).

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11415



Have verified that the patch fixes the specific failure originally reported. 
Simple file locking both with & without waits now appear to function
correctly.
 Haven''t been able to produce anything like the reported failure with
fcntl() or
flock() calls on 1 or 2 client nodes.  This isn''t an exhaustive test of
all
possible variations, just a few simple touch tests.

Looks good to me.