This is a case where there really is no good reason to run a shell command from within your perl script -- especially when the shell command is only going to run another perl script, as a one-liner. That's just way too complicated.

The easy way is to slurp the whole file into a scalar variable, do your regex substitution, and write the string back out to a file. (I'd recommend keeping the original as-is and saving the altered data to a different file, so you can be sure it worked before obliterating the original data.)

Here's one way to do your task:

#!/usr/bin/perl

use strict;

my $find = '(<!--\s*<control_link_address>push1\.mycompany\.com</contr
+ol_link_address>\s*-->)';
my $rplc = '$1\n <control_link_address>mdp.travel.co.uk</control_link_
+address>\n';

open( IN, "<", "/tmp/ls.conf" ) or die "/tmp/ls.conf: $!";
$/ = undef;
$_ = <IN>;
eval "s{$find}{$rplc}";
open( OUT, ">", "/tmp/ls.conf.new" ) or die "/tmp/ls.conf.new: $!";
print OUT;
[download]

I'm using a string eval so that the "$1" will behave as intended. I'm sure there are better ways... (Parsing, maybe?)