device entries outside /proc with procfs (for chroot)

Martin Cracauer cracauer at
Wed Sep 21 13:58:52 PDT 2005

Alexander Leidinger wrote on Tue, Sep 20, 2005 at 09:23:22PM +0200: 
> > It seemes that the controlled procfs mounting is the solution.  In my
> > case I don't chroot for security reasons, just to get the FreeBSD libs
> > and programs out of the way, so I don't even have to secure the second
> > mount.
> Yes, multiple devfs mounts are the way to go. Or mount linprocfs...

I have but it doesn't give me /dev/null :-)

> > What would be your idea of a proper Linux environment? They move
> > faster than I can follow :-)
> 8 is the default. If you don't have something which depends upon a
> newer one, use the default.

I am more concerned about older.

The thing is that a binary built on Redhat-7 works on 8, 9 and the
Fedora Cores (if those don't have their play-with-the-VM-map day).

By bumping it up you lose the ability to crosscompile for Rh-7 and its
derivates (RH enterprise Linux, whitebox) which are in wide use in
production environments.

Of course Redhat-7 had that "interesting" gcc-2.96 which I don't want
either so overall I'm happy with a RH-8 base.

> A lot of people use rh-9 (OVERRIDE_LINUX_BASE_PORT=rh9 in make.conf),
> but the port has some flaws and Trevor doesn't react. 

RH9/FC1 (same thing) sucks for the crosscompiler because the linker is
dead slow.  Linking a big C++ library is several times slower than RH7
or FC2, last time I looked (not properly benchmarked).

I also bet 45 cookies that moving past RH-9 breaks things for other

> I think I will
> claim a maintainer timeout soon (perhaps at the weekend if I get time)
> and fix some things (runtime linker path if you want to use the X11
> libs). I don't use it myself, but I haven't heard very bad things about
> it.

The current one works pretty well and doesn't seem to be a bad
compromise overall.

Then of course RH8 is among Linuxers known as the worst RH ever, so
what do I know?

Martin Cracauer <cracauer at>
FreeBSD - where you want to go, today.

More information about the freebsd-current mailing list