Shrink the xfs inode size by 68 bytes, add a new memory allocation
flag to indicate that disk I/O during page freeing within the
allocation path is OK. Where it is simple to figure out add this
to memory allocation calls, this hits the inode cache which is a
big user and the transaction structure itself.
Date: Mon Apr 16 15:49:09 PDT 2001
Workarea: jen.americas.sgi.com:/src/lord/xfs-base
The following file(s) were checked into:
bonnie.engr.sgi.com:/isms/slinx/2.4.x-xfs
Modid: 2.4.x-xfs:slinx:92653a
linux/fs/xfs/xfsidbg.c - 1.159
linux/fs/xfs/xfs_rw.c - 1.336
- remove references to deleted inode fields
linux/fs/xfs/xfs_da_btree.c - 1.118
- Add new memory flag to indicate IO during allocation is OK
linux/fs/xfs/xfs_iocore.c - 1.25
- remove references to deleted inode fields
linux/fs/xfs/xfs_iget.c - 1.137
linux/fs/xfs/xfs_mount.c - 1.253
linux/fs/xfs/xfs_inode.c - 1.316
- Add new memory flag to indicate IO during allocation is OK
linux/fs/xfs/xfs_inode.h - 1.146
- Prune out some inode fields we do not use, shrinks inode down 60 or
70 bytes.
linux/fs/xfs/xfs_dir2_leaf.c - 1.21
linux/fs/xfs/xfs_trans.c - 1.118
linux/fs/xfs_support/kmem.h - 1.3
- Add new memory flag to indicate IO during allocation is OK
linux/fs/xfs_support/kmem.c - 1.3
- Add new memory flag to indicate IO during allocation is OK
map this onto SLAB_KERNEL.
|