The latest installment of the ongoing saga of fixing up
inode locking. Previous changes re-arranged how we unlock
new inodes, to make sure that we set inode_ops before
we unlocked the inode. However, in a few cases that
was leaving the inode locked indefinately, which could cause
xfsdump (and probably others) to hang waiting for the inode.
This change adds a new function, xfs_iput_new, that handles
unlocking brand-new, locked inodes that we immediately decide
to release.
Date: Wed May 22 09:30:23 PDT 2002
Workarea:
stout.americas.sgi.com:/localhome/src/sandeen/2.4.x-xfs/workarea-testing
The following file(s) were checked into:
bonnie.engr.sgi.com:/isms/slinx/2.4.x-xfs
Modid: 2.4.x-xfs:slinx:119875a
linux/fs/xfs/xfs_itable.c - 1.103
linux/fs/xfs/xfs_dmapi.c - 1.52
linux/fs/xfs/xfs_qm_syscalls.c - 1.60
linux/fs/xfs/xfs_vfsops.c - 1.344
- Use xfs_iput_new for new inodes that are still locked
linux/fs/xfs/xfs_iget.c - 1.156
- New xfs_iput_new routine to handle inodes that are still locked
Use xfs_iput_new for new inodes that are still locked
linux/fs/xfs/xfs_qm.c - 1.73
- Use xfs_iput_new for new inodes that are still locked
linux/fs/xfs/xfs_inode.h - 1.159
- Prototype for xfs_iget_new
linux/fs/xfs/xfs_utils.c - 1.42
- Use xfs_iput_new for new inodes that are still locked
(Previously had some ingredients of xfs_iput in these
places, without actually calling xfs_iput).
linux/fs/xfs/linux/xfs_ioctl.c - 1.60
- Use xfs_iput_new for new inodes that are still locked
|