[Bug rtl-optimization/89487] [8/9 Regression] ICE in expand_expr_addr_expr_1, at expr.c:7993

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89487

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Can we ignore the DECL_HARD_REGISTER drs, or never canonicalize them to &var
(after all, they should never have any non-zero step, offset etc.), something
else?