Regarding this:

    open (CFGFILE, "ship.cfg") || die $!;
    while (my @TMP=<CFGFILE>) {
        next if ( /#/ );
        # Lets untaint the data after making sure it only matches
        # word characters... For more info read perldoc perlsec.
        foreach $_(@TMP) {
            if ($_ =~ /^([\w.]+)$/) {
                $_ = $1; 
                push (@hosts, $1);
            } else {
                next;
            }
        }
    }
[download]

That's totally wrong. Ordinarily, you either read from a file a line at a time, using while, or you read the entire (rest of the) file by assigning to a list. You can't meaningfully combine the two approaches, which is what it looks like you've tried to do. Below is how to write your loop using the one-fell-swoop approach, with the other transformations included:

    open CFGFILE, "< ship.cfg" or die "read ship.cfg: $!";
    @hosts =
        map { /^([\w.]+)$/ ? $1 : () }
        grep !/#/,
        <CFGFILE>;
    }
    close CFGFILE;
[download]

Here's something similar but using a line-at-a-time approach:

    open CFGFILE, "< ship.cfg" or die "read ship.cfg: $!";
    while (<CFGFILE>) {
        /#/ and next;
        /^([\w.]+)$/ or next;
        push @hosts, $1;
    }
    close CFGFILE;
[download]

hth.