From: John Tobin (tobinjt at domain netsoc.tcd.ie)
Date: Tue 23 Jul 2002 - 14:46:06 IST
On Sun, Jul 21, 2002 at 01:47:39AM +0100, Paul Jakma wrote:
> On Sat, 20 Jul 2002, kevin lyda wrote:
>
> anyway, links are directory entries that reference the same inode.
> (and i'll bet the restriction on directories is an artificial these
> days.. ELOOP exists anyway because of symlinks.)
Nope, it's not artificial. When you're dealing with symlinks you know
they're symlinks, so you can easily count them. E.g.:
$ mkdir foo; cd foo; ln -s ../foo bar; cd bar/bar/bar/bar . . .
The kernel can just increment the symlink count and return ELOOP if
necessary. But if you hardlink to a directory then the directory's
inode will tell you that it's a directory (which it is). You'll have
to cache the inodes traversed in reaching the current inode and search
the list each time, or take some similar approach to detect loops.
You could just not perform loop detection and rely on ENAMETOOLONG
though.
-- John "That would preempt a bunch of problems involved in trying to reconstruct exactly how the Perl 5 parser thinks, which nobody entirely understands." Larry Wall, 2001/04/20, perl6-language at domain perl.org "Finger to spiritual emptiness underlying everything." -- How a Japanese C manual referred to a "pointer to void".
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:18:01 GMT