G'day fidda,

This does what you want:

#!/usr/bin/env perl

use strict;
use warnings;

use Data::Dump;

my @A = qw{H1 H2 H3 H4};
my @B = 1 .. 10;

my %result;
my $reverse = 0;

while (@B) {
    build_hash(\%result, \@A, \@B, $reverse);
    $reverse ^= 1;
}

dd \%result;

sub build_hash {
    my ($result, $A, $B, $reverse) = @_;

    my @keys = @$B >= @$A ? @$A : @$A[0 .. $#$B];
    my @values;

    if ($reverse) {
        unshift @values, shift @$B for (0 .. $#keys);
    }
    else {
        push @values, shift @$B for (0 .. $#keys);
    }

    push @{$result->{$keys[$_]}}, $values[$_] for 0 .. $#keys;
}
[download]

Output:

{ H1 => [1, 8, 9], H2 => [2, 7, 10], H3 => [3, 6], H4 => [4, 5] }
[download]

Note that this code destroys @B. If you need to keep it, make a copy and use that in the while condition and when calling &build_hash:

my @B_copy = @B;
...
while (@B_copy) {
    build_hash(\%result, \@A, \@B_copy, $reverse);
...
[download]

Furthermore, the subroutine &build_hash has no reliance on the initial values of either @A or @B. For example, these initial values:

my @A = qw{H1 H2 H3 H4 H5};
my @B = 1 .. 23;
[download]

produce this output:

{
  H1 => [1, 10, 11, 20, 21],
  H2 => [2, 9, 12, 19, 22],
  H3 => [3, 8, 13, 18, 23],
  H4 => [4, 7, 14, 17],
  H5 => [5, 6, 15, 16],
}
[download]