> 1. The ON word is of genuine PG, and further of PIE, origin
> 2. The ON word is a borrowing
> 3. The ON word does NOT mean "Carpathians"

If the earliest record of the name in ON isn't just a chance
resemblance (3), then maybe the sequence of events went like this: The
name was borrowed into PG, before the time of the first consonant
shift, from another IE language where it was perhaps invented using a
PIE root [ http://en.wikipedia.org/wiki/Carpathian_Mountains ]. It
survived in at least some some dialect or dialects of East and/or West
Germanic until it became attached to stories about conflict between
the Goths and Huns. The name then travelled north with these stories
to Scandinavia (where it might have been known already, or might have
been known once but forgotten after contact was lost with the East
Germanic peoples who formerly inhabited the region; or maybe it was
never known in Scandinavia till it arrived with the legends).

Another name for mountains in the poem 'Jassar fjöll' has been
connected with German 'Gesenke', from Slavic 'Jeseník'
(Turville-Petre, p. 88). But I don't think this narrows down the date
it could have been borrowed at. Although initial /j/ inherited from
PG was lost c. 600 in Proto-Norse, /e/ underwent a later development
to /ja/ in certain circumstances (when followed in the next syllable
by a non-nasal /a/).

Regarding the origins of the material, Turville-Petre mentions a
couple of words used in the verses in meanings which are unusual for
ON, but match the meanings of their cognates in West Germanic
('skálkr' "servant", rather than "rogue"; 'skattr' "treasure", rather
then tribute), which "may be due to continental influence" (pp. 84-85,
p. 86).