> > I've observed a vfat problem; here's a session transcript:
> > /c is a vfat mount.
[cut]
> > when directories are moved, they become inaccessible. If I move it
> > back, it becomes accessible again. But only to the original directory.
> > Kernel 2.1.86.
> >
> > Can anyone verify this?
>
> Yep. This has been happening for as long as I can remember. It's a dcache
> thing. If you move the directory, unmount and remount the filesystem, it'll be
> where you put it.
The kernel didn't handle the hard links for vfat directories correctly, which
is shown by a simple stat call. So it thinks the moved directory is actually a
file. You can even cat it.
You may want to use the appended patch I got from Gordon Chaffee. This patch
is part of the kernel since 2.1.88, I think.
HTH,
Ulf
--- linux-2.1.85-clean/fs/vfat/namei.c Tue Jan 20 22:57:38 1998
+++ linux/fs/vfat/namei.c Fri Feb 13 08:19:03 1998
@@ -1761,12 +1768,15 @@
new_inode->i_mtime = old_inode->i_mtime;
new_inode->i_atime = old_inode->i_atime;
new_inode->i_ctime = old_inode->i_ctime;
+ new_inode->i_nlink = old_inode->i_nlink;
+ new_inode->i_op = old_inode->i_op;
MSDOS_I(new_inode)->i_ctime_ms = MSDOS_I(old_inode)->i_ctime_ms;
MSDOS_I(new_inode)->i_start = MSDOS_I(old_inode)->i_start;
MSDOS_I(new_inode)->i_logstart = MSDOS_I(old_inode)->i_logstart;
MSDOS_I(new_inode)->i_attrs = MSDOS_I(old_inode)->i_attrs;
+ fat_cache_inval_inode(old_inode);
mark_inode_dirty(new_inode);
old_dir->i_version = ++event;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu