You could load an AoA from your data file like this, which will retain the original ordering (if thats important), and allow you to sort the data pretty much any way you need to. Quite what you mean by

Data will be sorted on the d(n) field

is far from clear to me? Neither is why you think a "hashref" is relevant?

#! perl -slw
use strict;

my @data;
while( <DATA> ) {
    if( /^\s/ ) {
        push @{ $data[ -1 ] }, split;
    }
    else {
        push @data, [ split ];
    }
}

print "@$_" for @data;

=output
__DATA__
a1   b1   c1   d1   e1   f1
a2   b2   c2   d2   e2   f2
a3   b3   c3   d3   e3   f3
   p3   q3   r3
   s3   t3   u3
a4   b4   c4   d4   e4   f4
a5   b5   c5   d5   e5   f5
   p5   q5   r5
   s5   t5   u5
a6   b6   c6   d6   e6   f6
[download]

Output

C:\test>junk4
a1 b1 c1 d1 e1 f1
a2 b2 c2 d2 e2 f2
a3 b3 c3 d3 e3 f3 p3 q3 r3 s3 t3 u3
a4 b4 c4 d4 e4 f4
a5 b5 c5 d5 e5 f5 p5 q5 r5 s5 t5 u5
a6 b6 c6 d6 e6 f6
[download]