Re: [PATCH] root-hopping for pre-2.3.41-3

From: H. Peter Anvin (hpa@transmeta.com)
Date: Tue Jan 25 2000 - 19:26:14 EST


Followup to: <20000126011602.A1070@win.tue.nl>
By author: Guest section DW <dwguest@win.tue.nl>
In newsgroup: linux.dev.kernel
>
> and Peter answered
>
> > > Right... everyone on the old / move to the new /, although I guess
> > > that's technically not 100% necessary.
>
> but I wonder.
> Usually in such a situation not many processes will be around, but still..
> It would be strange if the interpretation of paths suddenly changed for
> processes other than the one doing this pivot_root.
> So, I would prefer these old processes all to remain where they are,
> with the same root inode (that just got a new name, namely /old_root).
>
> Thus, the only process that would get the new / as root would be the
> process doing this pivot_root() (and perhaps init?).
>

I think this really is the right solution... usually the process doing
this will *be* init. There are other things that need to be done,
too, like replacing the console fd's after the remount (otherwise you
will hold initrd busy by virtue of having /dev/console open.)

I therefore conclude that doing automatic stuff is pointless and
probably harmful. The recommended sequence for root-switching will
then look something like (error checking omitted):

        chdir("/oldroot");
        pivot_root("/newroot", "/oldroot");
        chroot(".");
        new_console = open("/dev/console", O_RDWR);
        dup2(new_console, 0);
        dup2(new_console, 1);
        dup2(new_console, 2);
        close(new_console);

        execl("/sbin/init", "init", 0);

-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."

- 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/



This archive was generated by hypermail 2b29 : Mon Jan 31 2000 - 21:00:15 EST