This is a modified version of the code which was breaking XFS
journal on raid5. By reducing the number of cases where we
use a single buffer head on a page to the cases where the
page is being completely written and is aligned on a 4K disk
boundary, we are avoiding the problem. Log recovery on raid5
does work with this combination. Most filesystem I/O on raid5
will be able to use the speedup.
Date: Fri Jul 27 15:39:54 PDT 2001
Workarea: 128.162.187.49:/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:99721a
linux/fs/pagebuf/page_buf.c - 1.96
- Turn back on the md optimization for doing larger writes, only allow
completely aligned pages.
|