TAKE - yet more dbench triggered fixes

[Steve Lord <lord@xxxxxxx>]

Looks like this is the real culprit, there was a code path which
memory pressure would find which caused a leak on the buffer head
count on a page. End result, pages which could never be reclaimed
and a slow spiral down into memory deadlock.

Date:  Wed Dec 12 13:54:42 PST 2001
Workarea:  jen.americas.sgi.com:/src/lord/xfs-linux.2.4

The following file(s) were checked into:
  bonnie.engr.sgi.com:/isms/slinx/2.4.x-xfs


Modid:  2.4.x-xfs:slinx:108339a
linux/fs/buffer.c - 1.98
        - Fix buffer count leak and clean up code a little