In 5.8 you can (and should) specify which encoding perl should use when writing.
By default perl5.8 will assume you want latin1 output (but warn and then output in utf8 those strings that do not fit in latin1).

If you switch your environment to utf8 with something like

env LC_CTYPE=en_US.UTF-8 perl -e ...
[download]

then you should get the output you expected.

A different way to do it (TIMTOWTDI) is to explicitly switch STDOUT to a different encoding using the newly repurposed binmode():

binmode STDOUT, ':utf8';
[download]

A third way is using the new use open pragma:

use open ':utf8', ':std';
[download]

So to get your one-liner to work, it becomes:

perl -Mopen=:utf8,:std -e '$c="f6"; $u=pack("U",hex($c)); print $u, "\
+x{f6}"' | od -tx1
[download]

I'm sure you'll be able to find your way to the documentation for these new perl5.8 features. A good start is the perluniintro manpage.