in fact, your code shouldn't be some thing like
    my $pattern = qr{ab[crdnfqm]def};
        if $string =~ m{ ($pattern) };
                 ....
which doesn't find all strings with the pattern.

That is perhaps because the regex  m{ ($pattern) } (which lacks the  /x regex modifier) requires a space before and after  $pattern in order to match.

I don't understand trailing xms at the end of pattern.

These are regex modifiers. The  /x modifier makes the regex compiler ignore whitepace (for improved readability). The  /m and  /s modifiers control the behavior of, respectively, the  ^ $ (string end) and  . (dot) metacharacter operators. I have adopted the best practice recommended by TheDamian of always (well, almost always) including these modifiers in every  qr// m// s/// regex I write. The reason for this is to improve regex readability (/x) and to make explicit the behavior of the  ^ $ . operators (/ms).

Please see perlre, perlrequick, and perlretut.

How to add + and space to the construct? So instead of

 my $pattern = qr{ab[crdnfqm]def};

 my $pattern = qr{ab[crdnfqm\+\s]def};
[download]

will be correct? to be used with your solution

 if $string =~ m{ ($pattern) };
[download]

The regex definition  qr{ab[crdnfqm\+\s]def}; does add a  '+' character and the class of all whitespace characters to the  [crdnfqm] character class contained in the previous  qr{ab[crdnfqm]def}; definition. Note, however, that  m{ ($pattern) } (still) requires a blank (0x20) character before and after  $pattern in order to have a match. (NB: The  '+' character is not special, i.e., is not a metacharacter, inside a character class and so does not need the  \ escape — but it does no harm.)

Please see perlre, perlrecharclass, perlrequick, and perlretut.