> viro@math.psu.edu (Alexander Viro) wrote on 14.11.99 in <Pine.GSO.4.10.9911141221440.2939-100000@weyl.math.psu.edu>:
>
> > On 14 Nov 1999, Kai Henningsen wrote:
> > > > 3.something. And it required the mounpoint to be immediate
> > > > subdirectory of root.
> > >
> > > Nope.
> >
> > Nope _what_? 3.30 definitely had JOIN, JOIN.EXE _did_ check for the depth
> > of mountpoint (want a reverse-engineered C source?) and while you could
> > do it by hands it was _bad_ idea. Reason: the same table kept the cwd for
> > each disk. It didn't take much to trick chdir() into buggering the whole
> > thing big way.
>
> DOS didn't always check the relevant length limits. That's a bug, not a
> missing feature.
Sure thing, that's a bug. But if you will consider the fact that the only
published interface to that table _did_ explicit check for the mountpoint
depth and refused to go if you were deeper than 1... And that was
documented. So, excuse me, but it sounds like "well, if they would clean
the chdir(2) up or change the data structures they could drop this check".
Well, duh. "If I tweak the /dev/kmem I can do everything, including
shooting my foot"...
> > Now, replace it with the setup when package foo sits in the directory
> > bar\baz and you bind x: to bar\baz. Setting all pathes in config as
> > absolute on x:
>
> Oh, you mean it gives you a very limited version of symlinks.
You never dealt with s5fs, right?[1] In some sense ~-expansion can also be
considered as a very limited version of symlinks...
[1] You really don't want to know the effect on one's sanity caused by mix
of NetWhore 2.<something>, DOS boxen, couple of Xenix 5 ones and <shudder>
VENIX.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/