From owner-cvs-sys  Fri Jun  2 04:01:18 1995
Return-Path: cvs-sys-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id EAA21215
          for cvs-sys-outgoing; Fri, 2 Jun 1995 04:01:18 -0700
Received: (from davidg@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id EAA21203
          ; Fri, 2 Jun 1995 04:01:09 -0700
Date: Fri, 2 Jun 1995 04:01:09 -0700
From: David Greenman <davidg>
Message-Id: <199506021101.EAA21203@freefall.cdrom.com>
To: CVS-commiters, cvs-sys
Subject: cvs commit: src/sys/msdosfs msdosfs_vnops.c
Sender: cvs-sys-owner@freebsd.org
Precedence: bulk

davidg      95/06/02 04:01:09

  Modified:    sys/msdosfs  msdosfs_vnops.c
  Log:
  Don't trash the denode cache entry for the source when renaming a
  directory in the newparent case.  For directories, the pseudo-inode
  number (de_dirclust, de_diroffset) is canonicalized to be
  (first cluster in directory, 0), so it doesn't change when directories
  are renamed.  This was only partly handled by not doing anything for
  directories in reinsert().
  
  It is hard to demonstrate bad effects of this bug.  You can see an
  unnecessary cache miss by tracing msdosfs_hashget().  The cache
  entry is usually left on the wrong hash chain so it is very
  unlikely to match anything, and bogus cache entries go away in
  msdosfs_reclaim().
  
  Submitted by:	Bruce Evans