linux.git/fs/omfs/dir.c, branch v3.2.90

omfs: fix (mode & S_IFDIR) abuse

2011-07-26T17:05:28+00:00

granted, on a filesystem that has only regular files and directories
it happens to work, but really should be S_ISDIR(mode)...

Signed-off-by: Al Viro

omfs: remove unnecessary dentry_unhash on rmdir, dir rneame

2011-05-28T05:02:52+00:00

omfs does not have problems with references to unlinked directories.

CC: Bob Copeland 
CC: linux-karma-devel@lists.sourceforge.net
Signed-off-by: Sage Weil 
Signed-off-by: Al Viro

vfs: push dentry_unhash on rename_dir into file systems

2011-05-26T11:26:48+00:00

Only a few file systems need this.  Start by pushing it down into each
rename method (except gfs2 and xfs) so that it can be dealt with on a
per-fs basis.

Acked-by: Christoph Hellwig 
Signed-off-by: Sage Weil 
Signed-off-by: Al Viro

vfs: push dentry_unhash on rmdir into file systems

2011-05-26T11:26:47+00:00

Only a few file systems need this.  Start by pushing it down into each
fs rmdir method (except gfs2 and xfs) so it can be dealt with on a per-fs
basis.

This does not change behavior for any in-tree file systems.

Acked-by: Christoph Hellwig 
Signed-off-by: Sage Weil 
Signed-off-by: Al Viro

omfs: make readdir stop when filldir says so

2011-03-05T21:24:12+00:00

filldir returning an error does *not* mean "skip this entry, try the
next one"...

Signed-off-by: Al Viro 
Signed-off-by: Bob Copeland

omfs: merge unlink() and rmdir(), close leak in rename()

2011-03-05T21:24:01+00:00

In case of directory-overwriting rename(), omfs forgot to mark the
victim doomed, so omfs_evict_inode() didn't free it.

We could fix that by calling omfs_rmdir() for directory victims
instead of doing omfs_unlink(), but it's easier to merge omfs_unlink()
and omfs_rmdir() instead.  Note that we have no hardlinks here.

It also makes the checks in omfs_rename() go away - they fold into
what omfs_remove() does when it runs into a directory.

Signed-off-by: Al Viro 
Signed-off-by: Bob Copeland

omfs: stop playing silly buggers with omfs_unlink() in ->rename()

2011-03-05T21:23:39+00:00

Since omfs directories are hashes of inodes and name is part of
inode, we have to remove inode from old directory before we can
put it into new one / under new name.  So instead of
	bump i_nlink
	call omfs_unlink, which does
		omfs_delete_entry()
		decrement i_nlink and mark parent dirty in case of success
	decrement i_nlink if omfs_unlink failed and hadn't done it itself
let's just call omfs_delete_entry() and dirty the parent ourselves...

Signed-off-by: Al Viro 
Signed-off-by: Bob Copeland

omfs: rename() needs to mark old_inode dirty after ctime update

2011-03-05T21:20:30+00:00

we *do* mark it dirty before, but it doesn't guarantee that we
don't get preempted just before assignment to ->i_ctime, with
inode getting written out before we get CPU back...

Signed-off-by: Al Viro 
Signed-off-by: Bob Copeland

omfs: check bounds on block numbers before passing to sb_bread

2010-07-10T18:37:06+00:00

In case of filesystem corruption, passing unchecked block numbers into
sb_bread can result in an infinite loop in __getblk().  Introduce a wrapper
function omfs_sbread() to check the block numbers and to also perform the
clus_to_blk() scaling.

Signed-off-by: Bob Copeland

const: constify remaining file_operations

2009-10-01T23:11:11+00:00

[akpm@linux-foundation.org: fix KVM]
Signed-off-by: Alexey Dobriyan 
Acked-by: Mike Frysinger 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds