Assume
$$ Lu=f\quad \text{in } [0,1]^d\\ u=0 \quad\text{ on } \partial[0,1]^d $$ for some strongly-elliptic operator $L$, and $f\in H^k$$, k\geq -1$. Do we have $u\in H^{k+2}$? I can only find the result for smooth domains.
I am pretty sure the answer is yes, for the Laplacian I can derive it myself. However, all proofs for general $L$ rely on "flattening the boundary", i.e. a $C^{k+2}$ boundary.