Re: Proposal: restrict link(2)

Harald Koenig (koenig@tat.physik.uni-tuebingen.de)
Wed, 11 Dec 1996 18:36:51 +0100 (MET)

> > > why is Joe Random Cracker allowed to make a hard link to /etc/passwd at all
> > > (or to any other file not owed by him) ?
>
> Why not? The user can copy the file, so why shouldn't it be possible to
> make a (hard or soft) link to the file. Or the user is _not_ allowed to
> copy (=read) the file, but in that case the user can't read the link
> either. The user doesn't gain any additional power by linking the file,
> the permissible operations on the file are exactly the same as on the
> original link (except, of course, the ability to unlink the file from the
> directory it was linked to).
>
> Sorry, but I think this is a non-issue and a case of not understanding the
> Unix file system semantics.

I think I'm very well understanding Unix file system semantics.
maybe I should have asked "what's the real benefit or reason to allow..."
or similar instead.

I can only see drawbacks in allowing this (same as that SYSV allowing
Joe U. to change the owner of his own files to any other user)
giving a number of problem and risks and no real use at all.
which possible good application am I missing ?

> One point remains, though: By making a hard link to a file, Joe A. User
> can make a file exist on a file system much longer than the owner of the
> file might expect.

this is a major problem esp. if you're using quotas and even more if Jim B.
has to pay for disk space (very expensive at the csd of your university:(
and Joe A. keeps hard links of your files in his own directories...

also you can change access policies to files (and not only to a copy
of the file) using hard links when access to the original directory
isn't possible for user/group B but Joe U. creates a hard link to this
file in a directory tree with different access rights...

Harald

--
All SCSI disks will from now on                     ___       _____
be required to send an email notice                0--,|    /OOOOOOO\
24 hours prior to complete hardware failure!      <_/  /  /OOOOOOOOOOO\
                                                    \  \/OOOOOOOOOOOOOOO\
                                                      \ OOOOOOOOOOOOOOOOO|//
Harald Koenig,                                         \/\/\/\/\/\/\/\/\/
Inst.f.Theoret.Astrophysik                              //  /     \\  \
koenig@tat.physik.uni-tuebingen.de                     ^^^^^       ^^^^^