A hole in our logic meant we were not cleaning dirty state on some
pages, end result was we wrote them twice - xfs was half the speed
of ext2 on streaming writes. With this change we are disk bound
again and back to ext2 speed. This is a 2.5 only change.
Date: Thu Sep 26 07:54:36 PDT 2002
Workarea: jen.americas.sgi.com:/src/lord/xfs-merge.2.5
The following file(s) were checked into:
bonnie.engr.sgi.com:/isms/slinx/2.5.x-xfs
Modid: 2.5.x-xfs:slinx:128467a
linux/fs/xfs/linux/xfs_aops.c - 1.9
- When we start I/O on a page, clean it, we need to do this ourselves
since we have a writepage call which writes more than one page out.
Also skip pages which are already being writeback processed somewhere.
|