I don't think you need the longest common substring, but the longest common tail of dirs.

here some code, it basically starts from the end and stops as soon as not all dirs are identical.

(can be done far more elegantly but lost interest ;-)

HTH, TIMTOWTDI!

use strict;
use warnings;

use Data::Dump qw(pp);

my @paths = map { [ split '/' ]  } qw[
        /home/instance/domains/abcd/xyz/1.txt
        /a1234/domains/abcd/xyz/1.txt
        /a1234/topfolder/instance/domains/abcd/xyz/1.txt
         ];
#pp @paths;

my $a_norm = shift @paths;

my @result;
 
dir_loop:
for my $idx ( 1 .. @$a_norm ) {
  for my $a_path (@paths){
    last dir_loop 
      if $a_norm->[-$idx] ne $a_path->[-$idx];
  } 
  unshift @result,$a_norm->[-$idx]
}
 
pp @result;
[download]

output:

("domains", "abcd", "xyz", "1.txt")
[download]