I wonder if this kenning, 'dróttinn hárar hallar foldar', would be a
counterexample to Ilya V. Sverdlov's claim that declined adjectives
are never attached to a non-root base word in kennings.

Kenning Morphology: Towards a Formal Definition of the Skaldic
Kenning, or Kennings and Adjectives
http://www.dur.ac.uk/medieval.www/sagaconf/sverdlov.htm

"If we could find examples of an agreed adjective attached to a
non-root base-word, or worse, to the determinant of a kenning, we
would be obliged to modify our proposition; but so far, there seem not
to be any."

He likens Old Norse kennings to German compound words in this respect:

"The real proof that this -s- is empty in both cases (Leben-s-art
versus Konstruktion-s-muster) is that we cannot attach an agreed
adjective to either Leben or Konstruktion when they appear as they do
in these compounds (see [v.1, 229]), while we can do so in
expressions like Art des Lebens. [...] And the adjectives that appear
to be attached to the root of the compound are, in fact, modifiers for
the whole compound, as their semantics readily testify."