pagebuf_iozero handles zeroing for xfs_zero_eof and xfs_zero_last_block
(both deal with zeroing unwritten portions when a write extends filesize).
But pagebuf_iozero wasn't setting up pages correctly, it was calling
pagebuf_commit_write_core w/o first setting up buffers on pages with
__pb_block_prepare_write_async.
xfs_zero_eof and xfs_zero_last_block were also forcing writes on these
pagebufs, which was probably causing double i/o...
Date: Wed Feb 20 13:27:57 PST 2002
Workarea: stout.americas.sgi.com:/localhome/eric/2.4.x-xfs/workarea-reallyclean
The following file(s) were checked into:
bonnie.engr.sgi.com:/isms/slinx/2.4.x-xfs
Modid: 2.4.x-xfs:slinx:112227a
linux/fs/xfs/linux/xfs_lrw.c - 1.123
- Prepare pagebuf bmap to pass to pagebuf_iozero
Don't force writing the pagebuf
linux/fs/xfs/pagebuf/page_buf_io.c - 1.13
- do a __pb_block_prepare_write_async before pagebuf_commit_write_core
linux/fs/xfs/pagebuf/page_buf.h - 1.4
- Add pb_bmap_t to pagebuf_iozero arguments
|